Mercurial > repos > mheinzl > fsd_regions

annotate fsd_regions.py @ 1:9ce2b4089c1b draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit b9403b3ce2b7a41fa8ee1aa47909152de78cf641
author mheinzl
date Tue, 15 May 2018 13:50:29 -0400
parents b82fdb006304
children 2631864873d7
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
1 #!/usr/bin/env python
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
2
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
3 # Family size distribution of tags which were aligned to the reference genome
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
4 #
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
5 # Author: Monika Heinzl, Johannes-Kepler University Linz (Austria)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
6 # Contact: monika.heinzl@edumail.at
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
7 #
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
8 # Takes at least one TABULAR file with tags before the alignment to the SSCS
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
9 # and a TXT with tags of reads that overlap the regions of the reference genome as input.
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
10 # The program produces a plot which shows the distribution of family sizes of the tags from the input files and
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
11 # a CSV file with the data of the plot.
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
12
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
13 # USAGE: python FSD_regions_1.6_FINAL.py --inputFile filenameSSCS --inputName1 filenameSSCS --ref_genome filenameRefGenome --sep "characterWhichSeparatesCSVFile" --output_csv outptufile_name_csv --output_pdf outptufile_name_pdf
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
14
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
15 import numpy
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
16 import matplotlib.pyplot as plt
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
17 import argparse
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
18 import sys
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
19 import os
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
20 from matplotlib.backends.backend_pdf import PdfPages
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
21
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
22 def readFileReferenceFree(file, delim):
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
23 with open(file, 'r') as dest_f:
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
24 data_array = numpy.genfromtxt(dest_f, skip_header=0, delimiter=delim, comments='#', dtype='string')
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
25 return(data_array)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
26
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
27 def make_argparser():
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
28 parser = argparse.ArgumentParser(description='Family Size Distribution of tags which were aligned to regions of the reference genome')
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
29 parser.add_argument('--inputFile',
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
30 help='Tabular File with three columns: ab or ba, tag and family size.')
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
31 parser.add_argument('--inputName1')
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
32 parser.add_argument('--ref_genome',
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
33 help='TXT File with tags of reads that overlap the region.')
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
34 parser.add_argument('--output_csv', default="data.csv", type=str,
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
35 help='Name of the pdf and csv file.')
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
36 parser.add_argument('--output_pdf', default="data.pdf", type=str,
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
37 help='Name of the pdf and csv file.')
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
38 parser.add_argument('--sep', default=",",
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
39 help='Separator in the csv file.')
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
40 return parser
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
41
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
42 def compare_read_families_refGenome(argv):
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
43 parser = make_argparser()
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
44 args = parser.parse_args(argv[1:])
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
45
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
46 firstFile = args.inputFile
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
47 name1 = args.inputName1
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
48 name1 = name1.split(".tabular")[0]
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
49 refGenome = args.ref_genome
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
50 title_file = args.output_pdf
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
51 title_file2 = args.output_csv
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
52 sep = args.sep
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
53
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
54 if type(sep) is not str or len(sep) > 1:
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
55 print("Error: --sep must be a single character.")
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
56 exit(3)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
57
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
58 with open(title_file2, "w") as output_file, PdfPages(title_file) as pdf:
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
59 data_array = readFileReferenceFree(firstFile, "\t")
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
60
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
61 mut_array = readFileReferenceFree(refGenome, " ")
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
62 length_regions = len(mut_array)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
63
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
64 seq = numpy.array(data_array[:, 1])
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
65 tags = numpy.array(data_array[:, 2])
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
66 quant = numpy.array(data_array[:, 0]).astype(int)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
67 group = numpy.array(mut_array[:, 0])
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
68
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
69 all_ab = seq[numpy.where(tags == "ab")[0]]
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
70 all_ba = seq[numpy.where(tags == "ba")[0]]
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
71 quant_ab = quant[numpy.where(tags == "ab")[0]]
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
72 quant_ba = quant[numpy.where(tags == "ba")[0]]
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
73
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
74 seqDic_ab = dict(zip(all_ab, quant_ab))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
75 seqDic_ba = dict(zip(all_ba, quant_ba))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
76
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
77 seq_mut = numpy.array(mut_array[:, 1])
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
78
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
79 groupUnique, group_index = numpy.unique(group, return_index=True)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
80 groupUnique = groupUnique[numpy.argsort(group_index)]
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
81
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
82 lst_ab = []
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
83 for i in seq_mut:
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
84 lst_ab.append(seqDic_ab.get(i))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
85
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
86 lst_ba = []
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
87 for i in seq_mut:
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
88 lst_ba.append(seqDic_ba.get(i))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
89
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
90 quant_ab = numpy.array(lst_ab)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
91 quant_ba = numpy.array(lst_ba)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
92
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
93 quantAfterRegion = []
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
94 for i in groupUnique:
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
95 dataAB = quant_ab[numpy.where(group == i)[0]]
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
96 dataBA = quant_ba[numpy.where(group == i)[0]]
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
97 bigFamilies = numpy.where(dataAB > 20)[0]
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
98 dataAB[bigFamilies] = 22
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
99 bigFamilies = numpy.where(dataBA > 20)[0]
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
100 dataBA[bigFamilies] = 22
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
101
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
102 quantAll = numpy.concatenate((dataAB, dataBA))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
103 quantAfterRegion.append(quantAll)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
104
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
105 maximumX = numpy.amax(numpy.concatenate(quantAfterRegion))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
106 minimumX = numpy.amin(numpy.concatenate(quantAfterRegion))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
107
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
108 ### PLOT ###
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
109 plt.rc('figure', figsize=(11.69, 8.27)) # A4 format
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
110 plt.rcParams['axes.facecolor'] = "E0E0E0" # grey background color
1
9ce2b4089c1b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit b9403b3ce2b7a41fa8ee1aa47909152de78cf641
mheinzl
parents: 0
diff changeset
111 plt.rcParams['xtick.labelsize'] = 14
9ce2b4089c1b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit b9403b3ce2b7a41fa8ee1aa47909152de78cf641
mheinzl
parents: 0
diff changeset
112 plt.rcParams['ytick.labelsize'] = 14
0
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
113 plt.rcParams['patch.edgecolor'] = "black"
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
114 fig = plt.figure()
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
115 plt.subplots_adjust(bottom=0.3)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
116
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
117 colors = ["#6E6E6E", "#0431B4", "#5FB404", "#B40431", "#F4FA58", "#DF7401", "#81DAF5"]
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
118
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
119 col = []
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
120 for i in range(0, len(groupUnique)):
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
121 col.append(colors[i])
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
122
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
123 counts = plt.hist(quantAfterRegion, bins=range(minimumX, maximumX + 1), stacked=False, label=groupUnique,
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
124 align="left", alpha=1, color=col, edgecolor="black", linewidth=1)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
125 ticks = numpy.arange(minimumX - 1, maximumX, 1)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
126
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
127 ticks1 = map(str, ticks)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
128 ticks1[len(ticks1) - 1] = ">20"
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
129 plt.xticks(numpy.array(ticks), ticks1)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
130 count = numpy.bincount(map(int, quant_ab)) # original counts
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
131
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
132 legend = "max. family size =\nabsolute frequency=\nrelative frequency=\n\ntotal nr. of reads="
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
133 plt.text(0.15, 0.105, legend, size=11, transform=plt.gcf().transFigure)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
134
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
135 legend = "AB\n{}\n{}\n{:.5f}\n\n{:,}" \
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
136 .format(max(map(int, quant_ab)), count[len(count) - 1], float(count[len(count) - 1]) / sum(count),
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
137 sum(numpy.array(data_array[:, 0]).astype(int)))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
138 plt.text(0.35, 0.105, legend, size=11, transform=plt.gcf().transFigure)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
139
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
140 count2 = numpy.bincount(map(int, quant_ba)) # original counts
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
141
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
142 legend = "BA\n{}\n{}\n{:.5f}" \
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
143 .format(max(map(int, quant_ba)), count2[len(count2) - 1], float(count2[len(count2) - 1]) / sum(count2))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
144 plt.text(0.45, 0.15, legend, size=11, transform=plt.gcf().transFigure)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
145
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
146 legend1 = "total nr. of tags="
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
147 legend2 = "total numbers * \n{:,}".format(length_regions)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
148 plt.text(0.6, 0.2, legend1, size=11, transform=plt.gcf().transFigure)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
149 plt.text(0.75, 0.2, legend2, size=11, transform=plt.gcf().transFigure)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
150 legend4 = "* In the plot, both family sizes of the ab and ba strands were used.\nWhereas the total numbers indicate only the single count of the tags per region.\n"
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
151 plt.text(0.1, 0.02, legend4, size=11, transform=plt.gcf().transFigure)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
152
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
153 space = numpy.arange(0, len(groupUnique), 0.02)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
154 for i, s, count in zip(groupUnique, space, quantAfterRegion):
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
155 plt.text(0.6, 0.05 + s, "{}=\n".format(i), size=11, transform=plt.gcf().transFigure)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
156 plt.text(0.75, 0.05 + s, "{:,}\n".format(len(count) / 2), size=11, transform=plt.gcf().transFigure)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
157
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
158 plt.legend(loc='upper right', fontsize=14, bbox_to_anchor=(0.9, 1), frameon=True)
1
9ce2b4089c1b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit b9403b3ce2b7a41fa8ee1aa47909152de78cf641
mheinzl
parents: 0
diff changeset
159 #plt.title(name1, fontsize=14)
9ce2b4089c1b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit b9403b3ce2b7a41fa8ee1aa47909152de78cf641
mheinzl
parents: 0
diff changeset
160 plt.xlabel("Family size", fontsize=14)
9ce2b4089c1b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit b9403b3ce2b7a41fa8ee1aa47909152de78cf641
mheinzl
parents: 0
diff changeset
161 plt.ylabel("Absolute Frequency", fontsize=14)
0
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
162 plt.grid(b=True, which="major", color="#424242", linestyle=":")
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
163 plt.margins(0.01, None)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
164
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
165 # plt.savefig("{}_regions.pdf".format(title_file), bbox_inch="tight")
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
166 pdf.savefig(fig, bbox_inch="tight")
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
167 plt.close()
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
168
1
9ce2b4089c1b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit b9403b3ce2b7a41fa8ee1aa47909152de78cf641
mheinzl
parents: 0
diff changeset
169 output_file.write("Dataset:{}{}\n".format(sep, name1))
0
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
170 output_file.write("{}AB{}BA\n".format(sep, sep))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
171 output_file.write("max. family size:{}{}{}{}\n".format(sep, max(map(int, quant_ab)), sep, max(map(int, quant_ba))))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
172 output_file.write("absolute frequency:{}{}{}{}\n".format(sep, count[len(count) - 1], sep, count2[len(count2) - 1]))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
173 output_file.write("relative frequency:{}{:.3f}{}{:.3f}\n\n".format(sep, float(count[len(count) - 1]) / sum(count), sep, float(count2[len(count2) - 1]) / sum(count2)))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
174 output_file.write("total nr. of reads{}{}\n".format(sep, sum(numpy.array(data_array[:, 0]).astype(int))))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
175 output_file.write("\n\nValues from family size distribution\n")
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
176 output_file.write("{}".format(sep))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
177 for i in groupUnique:
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
178 output_file.write("{}{}".format(i,sep))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
179 output_file.write("\n")
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
180 j=0
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
181 for fs in counts[1][0:len(counts[1])-1]:
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
182 if fs == 21:
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
183 fs = ">20"
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
184 else:
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
185 fs = "={}".format(fs)
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
186 output_file.write("FS{}{}".format(fs,sep))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
187 for n in range(len(groupUnique)):
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
188 output_file.write("{}{}".format(int(counts[0][n][j]), sep))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
189 output_file.write("\n")
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
190 j+=1
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
191 output_file.write("sum{}".format(sep))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
192 for i in counts[0]:
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
193 output_file.write("{}{}".format(int(sum(i)), sep))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
194 output_file.write("\n")
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
195 output_file.write("\n\nIn the plot, both family sizes of the ab and ba strands were used.\nWhereas the total numbers indicate only the single count of the tags per region.\n")
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
196 output_file.write("Region{}total nr. of tags per region\n".format(sep))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
197 for i, count in zip(groupUnique, quantAfterRegion):
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
198 output_file.write("{}{}{}\n".format(i,sep,len(count) / 2))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
199 output_file.write("sum of tags{}{}\n".format(sep,length_regions))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
200
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
201 print("Files successfully created!")
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
202 #print("Files saved under {}.pdf and {}.csv in {}!".format(title_file, title_file, os.getcwd()))
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
203
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
204 if __name__ == '__main__':
b82fdb006304 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd_regions commit 6055f8c5c052f528ff85fb5e0d43b4500830637a
mheinzl
parents:
diff changeset
205 sys.exit(compare_read_families_refGenome(sys.argv))