Mercurial > repos > mheinzl > hd
annotate hd.py @ 10:69aa17354a6e draft
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f01678e9bfead9f9e1b54dd9ecf7141f057dd9de
| author | mheinzl |
|---|---|
| date | Tue, 15 May 2018 11:27:27 -0400 |
| parents | c81bc96bea1c |
| children | 7adc48c8a03d |
| rev | line source |
|---|---|
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1 #!/usr/bin/env python |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
2 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
3 # Hamming distance analysis of SSCSs |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
4 # |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
5 # Author: Monika Heinzl, Johannes-Kepler University Linz (Austria) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
6 # Contact: monika.heinzl@edumail.at |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
7 # |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
8 # Takes at least one TABULAR file with tags before the alignment to the SSCS and optionally a second TABULAR file as input. |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
9 # The program produces a plot which shows a histogram of Hamming distances separated after family sizes, |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
10 # a family size distribution separated after Hamming distances for all (sample_size=0) or a given sample of SSCSs or SSCSs, which form a DCS. |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
11 # In additon, the tool produces HD and FSD plots for the difference between the HDs of both parts of the tags and for the chimeric reads |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
12 # and finally a CSV file with the data of the plots. |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
13 # It is also possible to perform the HD analysis with shortened tags with given sizes as input. |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
14 # The tool can run on a certain number of processors, which can be defined by the user. |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
15 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
16 # USAGE: python HDnew6_1Plot_FINAL.py --inputFile filename --inputName1 filename --inputFile2 filename2 --inputName2 filename2 --sample_size int/0 --sep "characterWhichSeparatesCSVFile" / |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
17 # --only_DCS True --FamilySize3 True --subset_tag True --nproc int --output_csv outptufile_name_csv --output_pdf outptufile_name_pdf |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
18 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
19 import numpy |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
20 import itertools |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
21 import operator |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
22 import matplotlib.pyplot as plt |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
23 import os.path |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
24 import cPickle as pickle |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
25 from multiprocessing.pool import Pool |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
26 from functools import partial |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
27 import argparse |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
28 import sys |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
29 import os |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
30 from matplotlib.backends.backend_pdf import PdfPages |
|
1
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
31 from collections import Counter |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
32 |
|
1
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
33 def plotFSDwithHD2(familySizeList1,maximumXFS,minimumXFS, quant, |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
34 title_file1, subtitle, pdf, relative=False, diff = True): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
35 if diff is False: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
36 colors = ["#e6194b", "#3cb44b", "#ffe119", "#0082c8", "#f58231", "#911eb4"] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
37 labels = ["HD=1", "HD=2", "HD=3", "HD=4", "HD=5-8","HD>8"] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
38 else: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
39 colors = ["#93A6AB", "#403C14", "#731E41", "#BAB591", "#085B6F", "#E8AA35", "#726C66"] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
40 if relative is True: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
41 labels = ["d=0", "d=0.1", "d=0.2", "d=0.3", "d=0.4", "d=0.5-0.8", "d>0.8"] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
42 else: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
43 labels = ["d=0","d=1", "d=2", "d=3", "d=4", "d=5-8","d>8"] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
44 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
45 fig = plt.figure(figsize=(6, 7)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
46 ax = fig.add_subplot(111) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
47 plt.subplots_adjust(bottom=0.1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
48 p1 = numpy.bincount(numpy.concatenate((familySizeList1))) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
49 maximumY = numpy.amax(p1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
50 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
51 if len(range(minimumXFS, maximumXFS)) == 0: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
52 range1 = range(minimumXFS - 1, minimumXFS + 2) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
53 else: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
54 range1 = range(0, maximumXFS + 2) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
55 counts = plt.hist(familySizeList1, label=labels, |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
56 color=colors, stacked=True, |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
57 rwidth=0.8,alpha=1, align="left", |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
58 edgecolor="None",bins=range1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
59 plt.legend(loc='upper right', fontsize=14, frameon=True, bbox_to_anchor=(1.45, 1)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
60 |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
61 #plt.title(title_file1, fontsize=12) |
|
1
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
62 plt.suptitle(subtitle, y=1, x=0.5, fontsize=14) |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
63 plt.xlabel("Family size", fontsize=14) |
|
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
64 plt.ylabel("Absolute Frequency", fontsize=14) |
|
1
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
65 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
66 ticks = numpy.arange(0, maximumXFS + 1, 1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
67 ticks1 = map(str, ticks) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
68 if maximumXFS >= 20: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
69 ticks1[len(ticks1) - 1] = ">=20" |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
70 plt.xticks(numpy.array(ticks), ticks1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
71 [l.set_visible(False) for (i, l) in enumerate(ax.get_xticklabels()) if i % 5 != 0] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
72 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
73 plt.xlim((0, maximumXFS + 1)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
74 if len(numpy.concatenate(familySizeList1)) != 0: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
75 plt.ylim((0, max(numpy.bincount(numpy.concatenate(familySizeList1))) * 1.1)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
76 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
77 plt.ylim((0, maximumY * 1.2)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
78 legend = "\nmax. family size: \nabsolute frequency: \nrelative frequency: " |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
79 plt.text(0.15, -0.08, legend, size=12, transform=plt.gcf().transFigure) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
80 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
81 count = numpy.bincount(quant) # original counts |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
82 legend1 = "{}\n{}\n{:.5f}" \ |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
83 .format(max(quant), count[len(count) - 1], float(count[len(count) - 1]) / sum(count)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
84 plt.text(0.5, -0.08, legend1, size=12, transform=plt.gcf().transFigure) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
85 legend3 = "singletons\n{:,}\n{:.5f}".format(int(counts[0][len(counts[0]) - 1][1]), |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
86 float(counts[0][len(counts[0]) - 1][1]) / sum( |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
87 counts[0][len(counts[0]) - 1])) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
88 plt.text(0.7, -0.08, legend3, transform=plt.gcf().transFigure, size=12) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
89 plt.grid(b=True, which='major', color='#424242', linestyle=':') |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
90 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
91 pdf.savefig(fig, bbox_inches="tight") |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
92 plt.close("all") |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
93 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
94 def plotHDwithFSD(list1,maximumX,minimumX, subtitle, lenTags, title_file1,pdf, |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
95 xlabel,relative=False): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
96 if relative is True: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
97 step = 0.1 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
98 else: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
99 step = 1 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
100 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
101 fig = plt.figure(figsize=(6, 8)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
102 plt.subplots_adjust(bottom=0.1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
103 con_list1 = numpy.concatenate(list1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
104 p1 = numpy.array([v for k, v in sorted(Counter(con_list1).iteritems())]) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
105 maximumY = numpy.amax(p1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
106 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
107 if relative is True: # relative difference |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
108 bin1 = numpy.arange(-1, maximumX + 0.2, 0.1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
109 else: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
110 bin1 = maximumX + 1 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
111 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
112 counts = plt.hist(list1, bins=bin1, edgecolor='black', linewidth=1, |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
113 label=["FS=1", "FS=2", "FS=3", "FS=4", "FS=5-10", |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
114 "FS>10"], rwidth=0.8, |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
115 color=["#808080", "#FFFFCC", "#FFBF00", "#DF0101", "#0431B4", "#86B404"], |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
116 stacked=True, alpha=1, |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
117 align="left", |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
118 range=(0, maximumX + 1)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
119 plt.legend(loc='upper right', fontsize=14, frameon=True, bbox_to_anchor=(1.45, 1)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
120 bins = counts[1] # width of bins |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
121 counts = numpy.array(map(int, counts[0][5])) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
122 plt.suptitle(subtitle, y=1, x=0.5, fontsize=14) |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
123 # plt.title(title_file1, fontsize=12) |
|
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
124 plt.xlabel(xlabel, fontsize=14) |
|
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
125 plt.ylabel("Absolute Frequency", fontsize=14) |
|
1
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
126 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
127 plt.grid(b=True, which='major', color='#424242', linestyle=':') |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
128 plt.axis((minimumX - step, maximumX + step, 0, numpy.amax(counts) + sum(counts) * 0.1)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
129 plt.xticks(numpy.arange(0, maximumX + step, step)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
130 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
131 plt.ylim((0, maximumY * 1.2)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
132 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
133 bin_centers = -0.4 * numpy.diff(bins) + bins[:-1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
134 for x_label, label in zip(counts, bin_centers): # labels for values |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
135 if x_label == 0: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
136 continue |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
137 else: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
138 plt.annotate("{:,}\n{:.3f}".format(x_label, float(x_label) / sum(counts), 1), |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
139 xy=(label, x_label + len(con_list1) * 0.01), |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
140 xycoords="data", color="#000066",fontsize=10) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
141 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
142 legend = "sample size= {:,} against {:,}".format(sum(counts), lenTags) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
143 plt.text(0.14, -0.01, legend, size=12, transform=plt.gcf().transFigure) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
144 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
145 pdf.savefig(fig, bbox_inches="tight") |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
146 plt.close("all") |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
147 plt.clf() |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
148 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
149 def plotHDwithinSeq_Sum2(sum1, sum2,min_value, lenTags, title_file1, pdf): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
150 fig = plt.figure(figsize=(6, 8)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
151 plt.subplots_adjust(bottom=0.1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
152 |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
153 ham = [sum1, sum2,numpy.array(min_value)] # new hd within tags |
|
1
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
154 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
155 maximumX = numpy.amax(numpy.concatenate(ham)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
156 minimumX = numpy.amin(numpy.concatenate(ham)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
157 maximumY = numpy.amax(numpy.concatenate(map(lambda (x): numpy.bincount(x), ham))) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
158 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
159 if len(range(minimumX, maximumX)) == 0: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
160 range1 = minimumX |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
161 else: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
162 range1 = range(minimumX, maximumX + 2) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
163 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
164 counts = plt.hist(ham, align="left", rwidth=0.8, stacked=False, |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
165 label=[ "HD a", "HD b","HD a+b"], |
|
10
69aa17354a6e
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f01678e9bfead9f9e1b54dd9ecf7141f057dd9de
mheinzl
parents:
9
diff
changeset
|
166 bins=range1, color=[ "#58ACFA", "#FA5858","#585858"], edgecolor='black', linewidth=1) |
|
1
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
167 plt.legend(loc='upper right', fontsize=14, frameon=True, bbox_to_anchor=(1.55, 1)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
168 plt.suptitle('Hamming distances within tags', fontsize=14) |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
169 #plt.title(title_file1, fontsize=12) |
|
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
170 plt.xlabel("Hamming Distance", fontsize=14) |
|
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
171 plt.ylabel("Absolute Frequency", fontsize=14) |
|
1
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
172 plt.grid(b=True, which='major', color='#424242', linestyle=':') |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
173 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
174 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
175 plt.axis((minimumX - 1, maximumX + 1, 0, maximumY * 1.1)) |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
176 plt.xticks(numpy.arange(0, maximumX + 1, 1.0)) |
|
1
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
177 plt.ylim((0, maximumY * 1.1)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
178 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
179 legend = "sample size= {:,} against {:,}".format(len(ham[0]), lenTags, lenTags) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
180 plt.text(0.14, -0.01, legend, size=12, transform=plt.gcf().transFigure) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
181 pdf.savefig(fig, bbox_inches="tight") |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
182 plt.close("all") |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
183 plt.clf() |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
184 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
185 def createTableFSD2(list1, diff=True): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
186 selfAB = numpy.concatenate(list1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
187 uniqueFS = numpy.unique(selfAB) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
188 nr = numpy.arange(0, len(uniqueFS), 1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
189 if diff is False: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
190 count = numpy.zeros((len(uniqueFS), 6)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
191 else: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
192 count = numpy.zeros((len(uniqueFS), 7)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
193 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
194 state = 1 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
195 for i in list1: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
196 counts = list(Counter(i).items()) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
197 hd = [item[0] for item in counts] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
198 c = [item[1] for item in counts] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
199 table = numpy.column_stack((hd, c)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
200 if len(table) == 0: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
201 state = state + 1 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
202 continue |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
203 else: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
204 if state == 1: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
205 for i, l in zip(uniqueFS, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
206 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
207 if j[0] == uniqueFS[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
208 count[l, 0] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
209 if state == 2: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
210 for i, l in zip(uniqueFS, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
211 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
212 if j[0] == uniqueFS[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
213 count[l, 1] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
214 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
215 if state == 3: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
216 for i, l in zip(uniqueFS, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
217 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
218 if j[0] == uniqueFS[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
219 count[l, 2] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
220 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
221 if state == 4: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
222 for i, l in zip(uniqueFS, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
223 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
224 if j[0] == uniqueFS[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
225 count[l, 3] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
226 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
227 if state == 5: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
228 for i, l in zip(uniqueFS, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
229 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
230 if j[0] == uniqueFS[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
231 count[l, 4] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
232 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
233 if state == 6: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
234 for i, l in zip(uniqueFS, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
235 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
236 if j[0] == uniqueFS[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
237 count[l, 5] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
238 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
239 if state == 7: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
240 for i, l in zip(uniqueFS, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
241 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
242 if j[0] == uniqueFS[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
243 count[l, 6] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
244 state = state + 1 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
245 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
246 sumRow = count.sum(axis=1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
247 sumCol = count.sum(axis=0) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
248 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
249 uniqueFS = uniqueFS.astype(str) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
250 if uniqueFS[len(uniqueFS) - 1] == "20": |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
251 uniqueFS[len(uniqueFS) - 1] = ">20" |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
252 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
253 first = ["FS={}".format(i) for i in uniqueFS] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
254 final = numpy.column_stack((first, count, sumRow)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
255 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
256 return (final, sumCol) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
257 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
258 def createFileFSD2(summary, sumCol, overallSum, output_file, name,sep, rel=False, diff=True): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
259 output_file.write(name) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
260 output_file.write("\n") |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
261 if diff is False: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
262 output_file.write("{}HD=1{}HD=2{}HD=3{}HD=4{}HD=5-8{}HD>8{}sum{}\n".format(sep,sep,sep,sep,sep,sep,sep,sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
263 else: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
264 if rel is False: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
265 output_file.write("{}diff=0{}diff=1{}diff=2{}diff=3{}diff=4{}diff=5-8{}diff>8{}sum{}\n".format(sep,sep,sep,sep,sep,sep,sep,sep,sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
266 else: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
267 output_file.write("{}diff=0{}diff=0.1{}diff=0.2{}diff=0.3{}diff=0.4{}diff=0.5-0.8{}diff>0.8{}sum{}\n".format(sep,sep,sep,sep,sep,sep,sep,sep,sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
268 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
269 for item in summary: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
270 for nr in item: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
271 if "FS" not in nr and "diff" not in nr: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
272 nr = nr.astype(float) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
273 nr = nr.astype(int) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
274 output_file.write("{}{}".format(nr,sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
275 output_file.write("\n") |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
276 output_file.write("sum{}".format(sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
277 sumCol = map(int, sumCol) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
278 for el in sumCol: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
279 output_file.write("{}{}".format(el,sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
280 output_file.write("{}{}".format(overallSum.astype(int),sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
281 output_file.write("\n\n") |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
282 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
283 def createTableHD(list1, row_label): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
284 selfAB = numpy.concatenate(list1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
285 uniqueHD = numpy.unique(selfAB) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
286 nr = numpy.arange(0, len(uniqueHD), 1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
287 count = numpy.zeros((len(uniqueHD), 6)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
288 state = 1 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
289 for i in list1: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
290 counts = list(Counter(i).items()) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
291 hd = [item[0] for item in counts] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
292 c = [item[1] for item in counts] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
293 table = numpy.column_stack((hd, c)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
294 if len(table) == 0: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
295 state = state + 1 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
296 continue |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
297 else: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
298 if state == 1: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
299 for i, l in zip(uniqueHD, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
300 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
301 if j[0] == uniqueHD[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
302 count[l, 0] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
303 if state == 2: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
304 for i, l in zip(uniqueHD, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
305 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
306 if j[0] == uniqueHD[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
307 count[l, 1] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
308 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
309 if state == 3: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
310 for i, l in zip(uniqueHD, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
311 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
312 if j[0] == uniqueHD[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
313 count[l, 2] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
314 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
315 if state == 4: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
316 for i, l in zip(uniqueHD, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
317 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
318 if j[0] == uniqueHD[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
319 count[l, 3] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
320 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
321 if state == 5: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
322 for i, l in zip(uniqueHD, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
323 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
324 if j[0] == uniqueHD[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
325 count[l, 4] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
326 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
327 if state == 6: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
328 for i, l in zip(uniqueHD, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
329 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
330 if j[0] == uniqueHD[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
331 count[l, 5] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
332 state = state + 1 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
333 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
334 sumRow = count.sum(axis=1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
335 sumCol = count.sum(axis=0) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
336 first = ["{}{}".format(row_label,i) for i in uniqueHD] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
337 final = numpy.column_stack((first, count, sumRow)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
338 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
339 return (final, sumCol) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
340 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
341 def createTableHDwithTags(list1): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
342 selfAB = numpy.concatenate(list1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
343 uniqueHD = numpy.unique(selfAB) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
344 nr = numpy.arange(0, len(uniqueHD), 1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
345 count = numpy.zeros((len(uniqueHD), 3)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
346 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
347 state = 1 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
348 for i in list1: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
349 counts = list(Counter(i).items()) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
350 hd = [item[0] for item in counts] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
351 c = [item[1] for item in counts] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
352 table = numpy.column_stack((hd, c)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
353 if len(table) == 0: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
354 state = state + 1 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
355 continue |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
356 else: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
357 if state == 1: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
358 for i, l in zip(uniqueHD, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
359 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
360 if j[0] == uniqueHD[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
361 count[l, 0] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
362 if state == 2: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
363 for i, l in zip(uniqueHD, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
364 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
365 if j[0] == uniqueHD[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
366 count[l, 1] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
367 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
368 if state == 3: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
369 for i, l in zip(uniqueHD, nr): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
370 for j in table: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
371 if j[0] == uniqueHD[l]: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
372 count[l, 2] = j[1] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
373 state = state + 1 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
374 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
375 sumRow = count.sum(axis=1) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
376 sumCol = count.sum(axis=0) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
377 first = ["HD={}".format(i) for i in uniqueHD] |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
378 final = numpy.column_stack((first, count, sumRow)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
379 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
380 return (final, sumCol) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
381 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
382 def createFileHD(summary, sumCol, overallSum, output_file, name,sep): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
383 output_file.write(name) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
384 output_file.write("\n") |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
385 output_file.write("{}FS=1{}FS=2{}FS=3{}FS=4{}FS=5-10{}FS>10{}sum{}\n".format(sep,sep,sep,sep,sep,sep,sep,sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
386 for item in summary: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
387 for nr in item: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
388 if "HD" not in nr and "diff" not in nr: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
389 nr = nr.astype(float) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
390 nr = nr.astype(int) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
391 output_file.write("{}{}".format(nr,sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
392 output_file.write("\n") |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
393 output_file.write("sum{}".format(sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
394 sumCol = map(int, sumCol) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
395 for el in sumCol: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
396 output_file.write("{}{}".format(el,sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
397 output_file.write("{}{}".format(overallSum.astype(int),sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
398 output_file.write("\n\n") |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
399 |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
400 def createFileHDwithinTag(summary, sumCol, overallSum, output_file, name,sep): |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
401 output_file.write(name) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
402 output_file.write("\n") |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
403 output_file.write("{}HD a+b;HD a{}HD b{}sum{}\n".format(sep,sep,sep,sep)) |
|
1
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
404 for item in summary: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
405 for nr in item: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
406 if "HD" not in nr: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
407 nr = nr.astype(float) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
408 nr = nr.astype(int) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
409 output_file.write("{}{}".format(nr,sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
410 output_file.write("\n") |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
411 output_file.write("sum{}".format(sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
412 sumCol = map(int, sumCol) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
413 for el in sumCol: |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
414 output_file.write("{}{}".format(el,sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
415 output_file.write("{}{}".format(overallSum.astype(int),sep)) |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
416 output_file.write("\n\n") |
|
7414792e1cb8
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 90ee23e393deb06fa5c15e3778fa23c39a25f7ce
mheinzl
parents:
0
diff
changeset
|
417 |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
418 def hamming(array1, array2): |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
419 res = 99 * numpy.ones(len(array1)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
420 i = 0 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
421 array2 = numpy.unique(array2) # remove duplicate sequences to decrease running time |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
422 for a in array1: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
423 dist = numpy.array([sum(itertools.imap(operator.ne, a, b)) for b in array2]) # fastest |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
424 res[i] = numpy.amin(dist[dist > 0]) # pick min distance greater than zero |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
425 #print(i) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
426 i += 1 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
427 return res |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
428 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
429 def hamming_difference(array1, array2, mate_b): |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
430 array2 = numpy.unique(array2) # remove duplicate sequences to decrease running time |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
431 array1_half = numpy.array([i[0:(len(i)) / 2] for i in array1]) # mate1 part1 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
432 array1_half2 = numpy.array([i[len(i) / 2:len(i)] for i in array1]) # mate1 part 2 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
433 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
434 array2_half = numpy.array([i[0:(len(i)) / 2] for i in array2]) # mate2 part1 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
435 array2_half2 = numpy.array([i[len(i) / 2:len(i)] for i in array2]) # mate2 part2 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
436 |
|
8
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
437 #diff11 = 999 * numpy.ones(len(array2)) |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
438 #relativeDiffList = 999 * numpy.ones(len(array2)) |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
439 #ham1 = 999 * numpy.ones(len(array2)) |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
440 #ham2 = 999 * numpy.ones(len(array2)) |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
441 #min_valueList = 999 * numpy.ones(len(array2)) |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
442 #min_tagsList = 999 * numpy.ones(len(array2)) |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
443 #diff11_zeros = 999 * numpy.ones(len(array2)) |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
444 #min_tagsList_zeros = 999 * numpy.ones(len(array2)) |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
445 |
|
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
446 |
|
8
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
447 diff11 = [] |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
448 relativeDiffList = [] |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
449 ham1 = [] |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
450 ham2 = [] |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
451 min_valueList = [] |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
452 min_tagsList = [] |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
453 diff11_zeros = [] |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
454 min_tagsList_zeros = [] |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
455 i = 0 # counter, only used to see how many HDs of tags were already calculated |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
456 if mate_b is False: # HD calculation for all a's |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
457 half1_mate1 = array1_half |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
458 half2_mate1 = array1_half2 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
459 half1_mate2 = array2_half |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
460 half2_mate2 = array2_half2 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
461 elif mate_b is True: # HD calculation for all b's |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
462 half1_mate1 = array1_half2 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
463 half2_mate1 = array1_half |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
464 half1_mate2 = array2_half2 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
465 half2_mate2 = array2_half |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
466 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
467 for a, b, tag in zip(half1_mate1, half2_mate1, array1): |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
468 ## exclude identical tag from array2, to prevent comparison to itself |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
469 sameTag = numpy.where(array2 == tag) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
470 indexArray2 = numpy.arange(0, len(array2), 1) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
471 index_withoutSame = numpy.delete(indexArray2, sameTag) # delete identical tag from the data |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
472 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
473 # all tags without identical tag |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
474 array2_half_withoutSame = half1_mate2[index_withoutSame] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
475 array2_half2_withoutSame = half2_mate2[index_withoutSame] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
476 #array2_withoutSame = array2[index_withoutSame] # whole tag (=not splitted into 2 halfs) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
477 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
478 dist = numpy.array([sum(itertools.imap(operator.ne, a, c)) for c in |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
479 array2_half_withoutSame]) # calculate HD of "a" in the tag to all "a's" or "b" in the tag to all "b's" |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
480 min_index = numpy.where(dist == dist.min()) # get index of min HD |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
481 min_value = dist[min_index] # get minimum HDs |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
482 min_tag_half2 = array2_half2_withoutSame[min_index] # get all "b's" of the tag or all "a's" of the tag with minimum HD |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
483 #min_tag = array2_withoutSame[min_index] # get whole tag with min HD |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
484 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
485 dist2 = numpy.array([sum(itertools.imap(operator.ne, b, e)) for e in |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
486 min_tag_half2]) # calculate HD of "b" to all "b's" or "a" to all "a's" |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
487 for d_1, d_2 in zip(min_value, dist2): |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
488 if mate_b is True: # half2, corrects the variable of the HD from both halfs if it is a or b |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
489 d = d_2 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
490 d2 = d_1 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
491 else: # half1, corrects the variable of the HD from both halfs if it is a or b |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
492 d = d_1 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
493 d2 = d_2 |
|
8
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
494 min_valueList.append(d + d2) |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
495 min_tagsList.append(tag) |
|
9
c81bc96bea1c
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 772695137fbccb155da4cdd1e0492333b94d0b06
mheinzl
parents:
8
diff
changeset
|
496 ham1.append(d) |
|
c81bc96bea1c
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 772695137fbccb155da4cdd1e0492333b94d0b06
mheinzl
parents:
8
diff
changeset
|
497 ham2.append(d2) |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
498 difference1 = abs(d - d2) |
|
8
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
499 diff11.append(difference1) |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
500 rel_difference = round(float(difference1) / (d + d2), 1) |
|
8
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
501 relativeDiffList.append(rel_difference) |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
502 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
503 #### tags which have identical parts: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
504 if d == 0 or d2 == 0: |
|
8
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
505 min_tagsList_zeros.append(tag) |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
506 difference1_zeros = abs(d - d2) |
|
8
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
507 diff11_zeros.append(difference1_zeros) |
|
e2596a4e1c56
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 24e245e913368abc55281d3bf22b2e1b8d60d26a
mheinzl
parents:
7
diff
changeset
|
508 i += 1 |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
509 |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
510 #print(i) |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
511 diff11 = [st for st in diff11 if st != 999] |
|
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
512 ham1 = [st for st in ham1 if st != 999] |
|
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
513 ham2 = [st for st in ham2 if st != 999] |
|
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
514 min_valueList = [st for st in min_valueList if st != 999] |
|
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
515 min_tagsList = [st for st in min_tagsList if st != 999] |
|
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
516 relativeDiffList = [st for st in relativeDiffList if st != 999] |
|
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
517 diff11_zeros = [st for st in diff11_zeros if st != 999] |
|
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
518 min_tagsList_zeros = [st for st in min_tagsList_zeros if st != 999] |
|
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
519 |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
520 return ([diff11, ham1, ham2, min_valueList, min_tagsList, relativeDiffList, diff11_zeros, min_tagsList_zeros]) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
521 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
522 def readFileReferenceFree(file): |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
523 with open(file, 'r') as dest_f: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
524 data_array = numpy.genfromtxt(dest_f, skip_header=0, delimiter='\t', comments='#', dtype='string') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
525 integers = numpy.array(data_array[:, 0]).astype(int) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
526 return(integers, data_array) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
527 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
528 def hammingDistanceWithFS(quant, ham): |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
529 quant = numpy.asarray(quant) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
530 maximum = max(ham) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
531 minimum = min(ham) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
532 ham = numpy.asarray(ham) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
533 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
534 singletons = numpy.where(quant == 1)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
535 data = ham[singletons] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
536 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
537 hd2 = numpy.where(quant == 2)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
538 data2 = ham[hd2] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
539 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
540 hd3 = numpy.where(quant == 3)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
541 data3 = ham[hd3] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
542 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
543 hd4 = numpy.where(quant == 4)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
544 data4 = ham[hd4] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
545 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
546 hd5 = numpy.where((quant >= 5) & (quant <= 10))[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
547 data5 = ham[hd5] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
548 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
549 hd6 = numpy.where(quant > 10)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
550 data6 = ham[hd6] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
551 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
552 list1 = [data, data2, data3, data4, data5, data6] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
553 return(list1, maximum, minimum) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
554 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
555 def familySizeDistributionWithHD(fs, ham, diff=False, rel = True): |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
556 hammingDistances = numpy.unique(ham) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
557 fs = numpy.asarray(fs) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
558 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
559 ham = numpy.asarray(ham) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
560 bigFamilies2 = numpy.where(fs > 19)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
561 if len(bigFamilies2) != 0: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
562 fs[bigFamilies2] = 20 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
563 maximum = max(fs) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
564 minimum = min(fs) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
565 if diff is True: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
566 hd0 = numpy.where(ham == 0)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
567 data0 = fs[hd0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
568 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
569 if rel is True: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
570 hd1 = numpy.where(ham == 0.1)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
571 else: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
572 hd1 = numpy.where(ham == 1)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
573 data = fs[hd1] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
574 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
575 if rel is True: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
576 hd2 = numpy.where(ham == 0.2)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
577 else: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
578 hd2 = numpy.where(ham == 2)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
579 data2 = fs[hd2] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
580 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
581 if rel is True: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
582 hd3 = numpy.where(ham == 0.3)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
583 else: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
584 hd3 = numpy.where(ham == 3)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
585 data3 = fs[hd3] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
586 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
587 if rel is True: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
588 hd4 = numpy.where(ham == 0.4)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
589 else: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
590 hd4 = numpy.where(ham == 4)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
591 data4 = fs[hd4] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
592 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
593 if rel is True: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
594 hd5 = numpy.where((ham >= 0.5) & (ham <= 0.8))[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
595 else: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
596 hd5 = numpy.where((ham >= 5) & (ham <= 8))[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
597 data5 = fs[hd5] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
598 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
599 if rel is True: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
600 hd6 = numpy.where(ham > 0.8)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
601 else: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
602 hd6 = numpy.where(ham > 8)[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
603 data6 = fs[hd6] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
604 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
605 if diff is True: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
606 list1 = [data0,data, data2, data3, data4, data5, data6] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
607 else: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
608 list1 = [data, data2, data3, data4, data5, data6] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
609 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
610 return(list1, hammingDistances, maximum, minimum) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
611 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
612 def make_argparser(): |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
613 parser = argparse.ArgumentParser(description='Hamming distance analysis of duplex sequencing data') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
614 parser.add_argument('--inputFile', |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
615 help='Tabular File with three columns: ab or ba, tag and family size.') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
616 parser.add_argument('--inputName1') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
617 parser.add_argument('--inputFile2',default=None, |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
618 help='Tabular File with three columns: ab or ba, tag and family size.') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
619 parser.add_argument('--inputName2') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
620 parser.add_argument('--sample_size', default=1000,type=int, |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
621 help='Sample size of Hamming distance analysis.') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
622 parser.add_argument('--sep', default=",", |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
623 help='Separator in the csv file.') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
624 parser.add_argument('--subset_tag', default=0,type=int, |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
625 help='The tag is shortened to the given number.') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
626 parser.add_argument('--nproc', default=4,type=int, |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
627 help='The tool runs with the given number of processors.') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
628 parser.add_argument('--only_DCS', action="store_false", # default=False, type=bool, |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
629 help='Only tags of the DCSs are included in the HD analysis') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
630 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
631 parser.add_argument('--minFS', default=1, type=int, |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
632 help='Only tags, which have a family size greater or equal than specified, are included in the HD analysis') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
633 parser.add_argument('--maxFS', default=0, type=int, |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
634 help='Only tags, which have a family size smaller or equal than specified, are included in the HD analysis') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
635 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
636 parser.add_argument('--output_csv', default="data.csv", type=str, |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
637 help='Name of the csv file.') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
638 parser.add_argument('--output_pdf', default="data.pdf", type=str, |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
639 help='Name of the pdf file.') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
640 parser.add_argument('--output_pdf2', default="data2.pdf", type=str, |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
641 help='Name of the pdf file.') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
642 parser.add_argument('--output_csv2', default="data2.csv", type=str, |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
643 help='Name of the csv file.') |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
644 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
645 return parser |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
646 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
647 def Hamming_Distance_Analysis(argv): |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
648 parser = make_argparser() |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
649 args = parser.parse_args(argv[1:]) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
650 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
651 file1 = args.inputFile |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
652 name1 = args.inputName1 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
653 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
654 file2 = args.inputFile2 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
655 name2 = args.inputName2 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
656 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
657 index_size = args.sample_size |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
658 title_savedFile_pdf = args.output_pdf |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
659 title_savedFile_pdf2 = args.output_pdf2 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
660 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
661 title_savedFile_csv = args.output_csv |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
662 title_savedFile_csv2 = args.output_csv2 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
663 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
664 sep = args.sep |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
665 onlyDuplicates = args.only_DCS |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
666 minFS = args.minFS |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
667 maxFS = args.maxFS |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
668 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
669 subset = args.subset_tag |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
670 nproc = args.nproc |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
671 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
672 ### input checks |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
673 if index_size < 0: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
674 print("index_size is a negative integer.") |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
675 exit(2) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
676 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
677 if nproc <= 0: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
678 print("nproc is smaller or equal zero") |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
679 exit(3) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
680 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
681 if type(sep) is not str or len(sep)>1: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
682 print("sep must be a single character.") |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
683 exit(4) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
684 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
685 if subset < 0: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
686 print("subset_tag is smaller or equal zero.") |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
687 exit(5) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
688 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
689 ### PLOT ### |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
690 plt.rcParams['axes.facecolor'] = "E0E0E0" # grey background color |
|
10
69aa17354a6e
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f01678e9bfead9f9e1b54dd9ecf7141f057dd9de
mheinzl
parents:
9
diff
changeset
|
691 plt.rcParams['xtick.labelsize'] = 14 |
|
69aa17354a6e
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f01678e9bfead9f9e1b54dd9ecf7141f057dd9de
mheinzl
parents:
9
diff
changeset
|
692 plt.rcParams['ytick.labelsize'] = 14 |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
693 plt.rcParams['patch.edgecolor'] = "#000000" |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
694 plt.rc('figure', figsize=(11.69, 8.27)) # A4 format |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
695 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
696 if file2 != str(None): |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
697 files = [file1, file2] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
698 name1 = name1.split(".tabular")[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
699 name2 = name2.split(".tabular")[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
700 names = [name1, name2] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
701 pdf_files = [title_savedFile_pdf, title_savedFile_pdf2] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
702 csv_files = [title_savedFile_csv, title_savedFile_csv2] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
703 else: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
704 files = [file1] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
705 name1 = name1.split(".tabular")[0] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
706 names = [name1] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
707 pdf_files = [title_savedFile_pdf] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
708 csv_files = [title_savedFile_csv] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
709 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
710 for f, name_file, pdf_f, csv_f in zip(files, names, pdf_files, csv_files): |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
711 with open(csv_f, "w") as output_file, PdfPages(pdf_f) as pdf: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
712 print("dataset: ", name_file) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
713 integers, data_array = readFileReferenceFree(f) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
714 data_array = numpy.array(data_array) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
715 int_f = numpy.array(data_array[:, 0]).astype(int) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
716 data_array = data_array[numpy.where(int_f >= minFS)] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
717 integers = integers[integers >= minFS] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
718 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
719 # select family size for tags |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
720 if maxFS > 0: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
721 int_f2 = numpy.array(data_array[:, 0]).astype(int) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
722 data_array = data_array[numpy.where(int_f2 <= maxFS)] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
723 integers = integers[integers <= maxFS] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
724 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
725 print("min FS", min(integers)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
726 print("max FS", max(integers)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
727 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
728 tags = data_array[:, 2] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
729 seq = data_array[:, 1] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
730 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
731 if onlyDuplicates is True: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
732 # find all unique tags and get the indices for ALL tags, but only once |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
733 u, index_unique, c = numpy.unique(numpy.array(seq), return_counts=True, return_index=True) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
734 d = u[c > 1] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
735 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
736 # get family sizes, tag for duplicates |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
737 duplTags_double = integers[numpy.in1d(seq, d)] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
738 duplTags = duplTags_double[0::2] # ab of DCS |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
739 duplTagsBA = duplTags_double[1::2] # ba of DCS |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
740 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
741 duplTags_tag = tags[numpy.in1d(seq, d)][0::2] # ab |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
742 duplTags_seq = seq[numpy.in1d(seq, d)][0::2] # ab - tags |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
743 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
744 data_array = numpy.column_stack((duplTags, duplTags_seq)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
745 data_array = numpy.column_stack((data_array, duplTags_tag)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
746 integers = numpy.array(data_array[:, 0]).astype(int) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
747 print("DCS in whole dataset", len(data_array)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
748 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
749 ## HD analysis for a subset of the tag |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
750 if subset > 0: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
751 tag1 = numpy.array([i[0:(len(i)) / 2] for i in data_array[:, 1]]) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
752 tag2 = numpy.array([i[len(i) / 2:len(i)] for i in data_array[:, 1]]) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
753 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
754 flanking_region_float = float((len(tag1[0]) - subset)) / 2 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
755 flanking_region = int(flanking_region_float) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
756 if flanking_region_float % 2 == 0: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
757 tag1_shorten = numpy.array([i[flanking_region:len(i) - flanking_region] for i in tag1]) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
758 tag2_shorten = numpy.array([i[flanking_region:len(i) - flanking_region] for i in tag2]) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
759 else: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
760 flanking_region_rounded = int(round(flanking_region, 1)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
761 flanking_region_rounded_end = len(tag1[0]) - subset - flanking_region_rounded |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
762 tag1_shorten = numpy.array( |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
763 [i[flanking_region:len(i) - flanking_region_rounded_end] for i in tag1]) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
764 tag2_shorten = numpy.array( |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
765 [i[flanking_region:len(i) - flanking_region_rounded_end] for i in tag2]) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
766 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
767 data_array_tag = numpy.array([i + j for i, j in zip(tag1_shorten, tag2_shorten)]) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
768 data_array = numpy.column_stack((data_array[:, 0], data_array_tag, data_array[:, 2])) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
769 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
770 print("length of tag= ", len(data_array[0, 1])) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
771 # select sample: if no size given --> all vs. all comparison |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
772 if index_size == 0: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
773 result = numpy.arange(0, len(data_array), 1) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
774 else: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
775 result = numpy.random.choice(len(integers), size=index_size, |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
776 replace=False) # array of random sequences of size=index.size |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
777 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
778 # with open("index_result1_{}.pkl".format(app_f), "wb") as o: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
779 # pickle.dump(result, o, pickle.HIGHEST_PROTOCOL) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
780 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
781 # comparison random tags to whole dataset |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
782 result1 = data_array[result, 1] # random tags |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
783 result2 = data_array[:, 1] # all tags |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
784 print("size of the whole dataset= ", len(result2)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
785 print("sample size= ", len(result1)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
786 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
787 # HD analysis of whole tag |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
788 proc_pool = Pool(nproc) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
789 chunks_sample = numpy.array_split(result1, nproc) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
790 ham = proc_pool.map(partial(hamming, array2=result2), chunks_sample) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
791 proc_pool.close() |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
792 proc_pool.join() |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
793 ham = numpy.concatenate(ham).astype(int) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
794 # with open("HD_whole dataset_{}.txt".format(app_f), "w") as output_file1: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
795 # for h, tag in zip(ham, result1): |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
796 # output_file1.write("{}\t{}\n".format(tag, h)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
797 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
798 # HD analysis for chimeric reads |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
799 proc_pool_b = Pool(nproc) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
800 diff_list_a = proc_pool_b.map(partial(hamming_difference, array2=result2, mate_b=False), chunks_sample) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
801 diff_list_b = proc_pool_b.map(partial(hamming_difference, array2=result2, mate_b=True), chunks_sample) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
802 proc_pool_b.close() |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
803 proc_pool_b.join() |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
804 diff = numpy.concatenate((numpy.concatenate([item[0] for item in diff_list_a]), |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
805 numpy.concatenate([item_b[0] for item_b in diff_list_b]))).astype(int) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
806 HDhalf1 = numpy.concatenate((numpy.concatenate([item[1] for item in diff_list_a]), |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
807 numpy.concatenate([item_b[1] for item_b in diff_list_b]))).astype(int) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
808 HDhalf2 = numpy.concatenate((numpy.concatenate([item[2] for item in diff_list_a]), |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
809 numpy.concatenate([item_b[2] for item_b in diff_list_b]))).astype(int) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
810 minHDs = numpy.concatenate((numpy.concatenate([item[3] for item in diff_list_a]), |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
811 numpy.concatenate([item_b[3] for item_b in diff_list_b]))).astype(int) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
812 minHD_tags = numpy.concatenate((numpy.concatenate([item[4] for item in diff_list_a]), |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
813 numpy.concatenate([item_b[4] for item_b in diff_list_b]))) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
814 rel_Diff = numpy.concatenate((numpy.concatenate([item[5] for item in diff_list_a]), |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
815 numpy.concatenate([item_b[5] for item_b in diff_list_b]))) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
816 diff_zeros = numpy.concatenate((numpy.concatenate([item[6] for item in diff_list_a]), |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
817 numpy.concatenate([item_b[6] for item_b in diff_list_b]))).astype(int) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
818 minHD_tags_zeros = numpy.concatenate((numpy.concatenate([item[7] for item in diff_list_a]), |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
819 numpy.concatenate([item_b[7] for item_b in diff_list_b]))) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
820 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
821 # with open("HD_within tag_{}.txt".format(app_f), "w") as output_file2: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
822 # for d, s1, s2, hd, rel_d, tag in zip(diff, HDhalf1, HDhalf2, minHDs, rel_Diff, minHD_tags): |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
823 # output_file2.write( |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
824 # "{}\t{}\t{}\t{}\t{}\t{}\n".format(tag, hd, s1, s2, d, rel_d)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
825 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
826 lenTags = len(data_array) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
827 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
828 quant = numpy.array(data_array[result, 0]).astype(int) # family size for sample of tags |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
829 seq = numpy.array(data_array[result, 1]) # tags of sample |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
830 ham = numpy.asarray(ham) # HD for sample of tags |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
831 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
832 if onlyDuplicates is True: # ab and ba strands of DCSs |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
833 quant = numpy.concatenate((quant, duplTagsBA[result])) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
834 seq = numpy.tile(seq, 2) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
835 ham = numpy.tile(ham, 2) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
836 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
837 # prepare data for different kinds of plots |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
838 # distribution of FSs separated after HD |
|
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
839 familySizeList1, hammingDistances, maximumXFS, minimumXFS = familySizeDistributionWithHD(quant, ham,rel=False) |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
840 list1, maximumX, minimumX = hammingDistanceWithFS(quant, ham) # histogram of HDs separated after FS |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
841 |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
842 ## get FS for all tags with min HD of analysis of chimeric reads |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
843 # there are more tags than sample size in the plot, because one tag can have multiple minimas |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
844 seqDic = dict(zip(seq, quant)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
845 lst_minHD_tags = [] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
846 for i in minHD_tags: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
847 lst_minHD_tags.append(seqDic.get(i)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
848 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
849 # histogram with absolute and relative difference between HDs of both parts of the tag |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
850 listDifference1, maximumXDifference, minimumXDifference = hammingDistanceWithFS(lst_minHD_tags, diff) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
851 listRelDifference1, maximumXRelDifference, minimumXRelDifference = hammingDistanceWithFS(lst_minHD_tags, |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
852 rel_Diff) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
853 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
854 # family size distribution separated after the difference between HDs of both parts of the tag |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
855 familySizeList1_diff, hammingDistances_diff, maximumXFS_diff, minimumXFS_diff = familySizeDistributionWithHD( |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
856 lst_minHD_tags, diff, diff=True, rel=False) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
857 familySizeList1_reldiff, hammingDistances_reldiff, maximumXFS_reldiff, minimumXFS_reldiff = familySizeDistributionWithHD( |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
858 lst_minHD_tags, rel_Diff, diff=True, rel=True) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
859 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
860 # chimeric read analysis: tags which have HD=0 in one of the halfs |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
861 if len(minHD_tags_zeros) != 0: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
862 lst_minHD_tags_zeros = [] |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
863 for i in minHD_tags_zeros: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
864 lst_minHD_tags_zeros.append(seqDic.get(i)) # get family size for tags of chimeric reads |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
865 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
866 # histogram with HD of non-identical half |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
867 listDifference1_zeros, maximumXDifference_zeros, minimumXDifference_zeros = hammingDistanceWithFS( |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
868 lst_minHD_tags_zeros, diff_zeros) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
869 # family size distribution of non-identical half |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
870 familySizeList1_diff_zeros, hammingDistances_diff_zeros, maximumXFS_diff_zeros, minimumXFS_diff_zeros = familySizeDistributionWithHD( |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
871 lst_minHD_tags_zeros, diff_zeros, diff=False, rel=False) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
872 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
873 ########################## Plot HD within tags ######################################################## |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
874 ###################################################################################################################### |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
875 plotHDwithinSeq_Sum2(HDhalf1, HDhalf2, minHDs, pdf=pdf, lenTags=lenTags, title_file1=name_file) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
876 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
877 ##################################################################################################################### |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
878 ################## plot Hamming Distance with Family size distribution ############################## |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
879 ##################################################################################################################### |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
880 plotHDwithFSD(list1=list1, maximumX=maximumX, minimumX=minimumX, pdf=pdf, |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
881 subtitle="Hamming distance separated by family size", title_file1=name_file, |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
882 lenTags=lenTags,xlabel="Hamming distance") |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
883 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
884 ########################## Plot FSD with separation after HD ############################################### |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
885 ######################################################################################################################## |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
886 plotFSDwithHD2(familySizeList1, maximumXFS, minimumXFS, |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
887 quant=quant, subtitle="Family size distribution separated by Hamming distance", |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
888 pdf=pdf,relative=False, title_file1=name_file, diff=False) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
889 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
890 ########################## Plot difference between HD's separated after FSD ########################################## |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
891 ######################################################################################################################## |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
892 plotHDwithFSD(listDifference1, maximumXDifference, minimumXDifference, pdf=pdf, |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
893 subtitle="Delta Hamming distance within tags", |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
894 title_file1=name_file, lenTags=lenTags, |
|
10
69aa17354a6e
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f01678e9bfead9f9e1b54dd9ecf7141f057dd9de
mheinzl
parents:
9
diff
changeset
|
895 xlabel="absolute delta Hamming distance", relative=False) |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
896 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
897 plotHDwithFSD(listRelDifference1, maximumXRelDifference, minimumXRelDifference, pdf=pdf, |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
898 subtitle="Relative delta Hamming distances within tags", |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
899 title_file1=name_file, lenTags=lenTags, |
|
10
69aa17354a6e
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f01678e9bfead9f9e1b54dd9ecf7141f057dd9de
mheinzl
parents:
9
diff
changeset
|
900 xlabel="relative delta Hamming distance", relative=True) |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
901 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
902 #################### Plot FSD separated after difference between HD's ##################################### |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
903 ######################################################################################################################## |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
904 plotFSDwithHD2(familySizeList1_diff, maximumXFS_diff, minimumXFS_diff, |
|
10
69aa17354a6e
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f01678e9bfead9f9e1b54dd9ecf7141f057dd9de
mheinzl
parents:
9
diff
changeset
|
905 subtitle="Family size distribution separated by delta Hamming distances within the tags", |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
906 pdf=pdf,relative=False, diff=True, title_file1=name_file, quant=quant) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
907 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
908 plotFSDwithHD2(familySizeList1_reldiff, maximumXFS_reldiff, minimumXFS_reldiff, quant=quant, pdf=pdf, |
|
10
69aa17354a6e
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f01678e9bfead9f9e1b54dd9ecf7141f057dd9de
mheinzl
parents:
9
diff
changeset
|
909 subtitle="Family size distribution separated by delta Hamming distances within the tags", |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
910 relative=True, diff=True, title_file1=name_file) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
911 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
912 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
913 # plots for chimeric reads |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
914 if len(minHD_tags_zeros) != 0: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
915 ## HD |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
916 plotHDwithFSD(listDifference1_zeros, maximumXDifference_zeros, minimumXDifference_zeros, pdf=pdf, |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
917 subtitle="Hamming distance of the non-identical half of chimeras", |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
918 title_file1=name_file, lenTags=lenTags,xlabel="Hamming distance", relative=False) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
919 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
920 ## FSD |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
921 plotFSDwithHD2(familySizeList1_diff_zeros, maximumXFS_diff_zeros, minimumXFS_diff_zeros, |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
922 quant=quant, pdf=pdf, |
|
10
69aa17354a6e
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f01678e9bfead9f9e1b54dd9ecf7141f057dd9de
mheinzl
parents:
9
diff
changeset
|
923 subtitle="Family size distribution separated by Hamming distance of the non-identical half of chimeras", |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
924 relative=False, diff=False, title_file1=name_file) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
925 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
926 ### print all data to a CSV file |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
927 #### HD #### |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
928 summary, sumCol = createTableHD(list1, "HD=") |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
929 overallSum = sum(sumCol) # sum of columns in table |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
930 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
931 #### FSD #### |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
932 summary5, sumCol5 = createTableFSD2(familySizeList1, diff=False) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
933 overallSum5 = sum(sumCol5) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
934 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
935 ### HD of both parts of the tag #### |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
936 summary9, sumCol9 = createTableHDwithTags([numpy.array(minHDs), HDhalf1, HDhalf2]) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
937 overallSum9 = sum(sumCol9) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
938 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
939 ## HD |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
940 # absolute difference |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
941 summary11, sumCol11 = createTableHD(listDifference1, "diff=") |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
942 overallSum11 = sum(sumCol11) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
943 # relative difference and all tags |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
944 summary13, sumCol13 = createTableHD(listRelDifference1, "diff=") |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
945 overallSum13 = sum(sumCol13) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
946 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
947 ## FSD |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
948 # absolute difference |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
949 summary19, sumCol19 = createTableFSD2(familySizeList1_diff) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
950 overallSum19 = sum(sumCol19) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
951 # relative difference |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
952 summary21, sumCol21 = createTableFSD2(familySizeList1_reldiff) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
953 overallSum21 = sum(sumCol21) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
954 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
955 # chimeric reads |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
956 if len(minHD_tags_zeros) != 0: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
957 # absolute difference and tags where at least one half has HD=0 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
958 summary15, sumCol15 = createTableHD(listDifference1_zeros, "diff=") |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
959 overallSum15 = sum(sumCol15) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
960 # absolute difference and tags where at least one half has HD=0 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
961 summary23, sumCol23 = createTableFSD2(familySizeList1_diff_zeros, diff=False) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
962 overallSum23 = sum(sumCol23) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
963 |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
964 output_file.write("{}\n".format(name_file)) |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
965 output_file.write("number of tags per file{}{:,} (from {:,}) against {:,}\n\n".format(sep, len( |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
966 numpy.concatenate(list1)), lenTags, lenTags)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
967 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
968 ### HD ### |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
969 createFileHD(summary, sumCol, overallSum, output_file, |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
970 "Hamming distance separated by family size", sep) |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
971 ### FSD ### |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
972 createFileFSD2(summary5, sumCol5, overallSum5, output_file, |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
973 "Family size distribution separated by Hamming distance", sep, |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
974 diff=False) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
975 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
976 count = numpy.bincount(quant) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
977 output_file.write("{}{}\n".format(sep, f)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
978 output_file.write("max. family size:{}{}\n".format(sep, max(quant))) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
979 output_file.write("absolute frequency:{}{}\n".format(sep, count[len(count) - 1])) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
980 output_file.write( |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
981 "relative frequency:{}{}\n\n".format(sep, float(count[len(count) - 1]) / sum(count))) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
982 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
983 ### HD within tags ### |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
984 output_file.write( |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
985 "The hamming distances were calculated by comparing each half of all tags against the tag(s) with the minimum Hamming distance per half.\n" |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
986 "It is possible that one tag can have the minimum HD from multiple tags, so the sample size in this calculation differs from the sample size entered by the user.\n") |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
987 output_file.write( |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
988 "actual number of tags with min HD = {:,} (sample size by user = {:,})\n".format( |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
989 len(numpy.concatenate(listDifference1)), len(numpy.concatenate(list1)))) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
990 output_file.write("length of one part of the tag = {}\n\n".format(len(data_array[0, 1]) / 2)) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
991 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
992 createFileHDwithinTag(summary9, sumCol9, overallSum9, output_file, |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
993 "Hamming distance of each half in the tag", sep) |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
994 createFileHD(summary11, sumCol11, overallSum11, output_file, |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
995 "Absolute delta Hamming distances within the tag", sep) |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
996 createFileHD(summary13, sumCol13, overallSum13, output_file, |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
997 "Relative delta Hamming distances within the tag", sep) |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
998 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
999 createFileFSD2(summary19, sumCol19, overallSum19, output_file, |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
1000 "Family size distribution separated by absolute delta Hamming distance", |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1001 sep) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1002 createFileFSD2(summary21, sumCol21, overallSum21, output_file, |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
1003 "Family size distribution separated by relative delta Hamming distance", |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1004 sep, rel=True) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1005 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1006 if len(minHD_tags_zeros) != 0: |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1007 output_file.write( |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
1008 "Identifiaction of chimeric reads:\nAll tags were filtered: only those tags where at least one half is identical with the half of the min. tag are kept.\nSo the hamming distance of the non-identical half is compared.\n") |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1009 createFileHD(summary15, sumCol15, overallSum15, output_file, |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
1010 "Hamming distances of non-zero half", sep) |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1011 createFileFSD2(summary23, sumCol23, overallSum23, output_file, |
|
2
316fbf91dd12
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit f9d5547849dabb59a33a5e998bda4730323d62a9
mheinzl
parents:
1
diff
changeset
|
1012 "Family size distribution separated by Hamming distance of non-zero half", |
|
0
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1013 sep, diff=False) |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1014 output_file.write("\n") |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1015 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1016 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1017 |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1018 if __name__ == '__main__': |
|
239c4448a163
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/hd commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff
changeset
|
1019 sys.exit(Hamming_Distance_Analysis(sys.argv)) |