Mercurial > repos > fubar > bigwig_outlier_bed

annotate bigwig_outlier_bed.py @ 6:eb17eb8a3658 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
author fubar
date Tue, 23 Jul 2024 23:12:23 +0000
parents c71db540eb38
children c8e22efcaeda
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
1 """
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
2 Ross Lazarus June 2024 for VGP
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
3 Bigwigs are great, but hard to reliably "see" small low coverage or small very high coverage regions.
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
4 Colouring in JB2 tracks will need a new plugin, so this code will find bigwig regions above and below a chosen percentile point.
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
5 0.99 and 0.01 work well in testing with a minimum span of 10 bp.
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
6 Multiple bigwigs **with the same reference** can be combined - bed segments will be named appropriately
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
7 Combining multiple references works but is silly because only display will rely on one reference so others will not be shown...
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
8 Tricksy numpy method from http://gregoryzynda.com/python/numpy/contiguous/interval/2019/11/29/contiguous-regions.html
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
9 takes about 95 seconds for a 17MB test wiggle
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
10 JBrowse2 bed normally displays ignore the score, so could provide separate low/high bed file outputs as an option.
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
11 Update june 30 2024: wrote a 'no-build' plugin for beds to display red/blue if >0/<0 so those are used for scores
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
12 Bed interval naming must be short for JB2 but needs input bigwig name and (lo or hi).
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
13 """
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
14
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
15 import argparse
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
16 import os
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
17 import sys
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
18 from pathlib import Path
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
19
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
20 import numpy as np
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
21 import pybigtools
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
22
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
23
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
24 class asciihist:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
25
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
26 def __init__(
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
27 self,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
28 data,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
29 bins=10,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
30 minmax=None,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
31 str_tag="",
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
32 scale_output=80,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
33 generate_only=True,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
34 ):
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
35 """
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
36 https://gist.github.com/bgbg/608d9ef4fd75032731651257fe67fc81
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
37 Create an ASCII histogram from an interable of numbers.
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
38 Author: Boris Gorelik boris@gorelik.net. based on http://econpy.googlecode.com/svn/trunk/pytrix/pytrix.py
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
39 License: MIT
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
40 """
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
41 self.data = data
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
42 self.minmax = minmax
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
43 self.str_tag = str_tag
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
44 self.bins = bins
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
45 self.generate_only = generate_only
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
46 self.scale_output = scale_output
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
47 self.itarray = np.asanyarray(self.data)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
48 if self.minmax == "auto":
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
49 self.minmax = np.percentile(data, [5, 95])
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
50 if self.minmax[0] == self.minmax[1]:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
51 # for very ugly distributions
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
52 self.minmax = None
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
53 if self.minmax is not None:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
54 # discard values that are outside minmax range
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
55 mn = self.minmax[0]
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
56 mx = self.minmax[1]
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
57 self.itarray = self.itarray[self.itarray >= mn]
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
58 self.itarray = self.itarray[self.itarray <= mx]
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
59
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
60 def draw(self):
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
61 values, counts = np.unique(self.data, return_counts=True)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
62 if len(values) <= 20:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
63 self.bins = len(values)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
64 ret = []
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
65 if self.itarray.size:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
66 total = len(self.itarray)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
67 counts, cutoffs = np.histogram(self.itarray, bins=self.bins)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
68 cutoffs = cutoffs[1:]
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
69 if self.str_tag:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
70 self.str_tag = "%s " % self.str_tag
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
71 else:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
72 self.str_tag = ""
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
73 if self.scale_output is not None:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
74 scaled_counts = counts.astype(float) / counts.sum() * self.scale_output
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
75 else:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
76 scaled_counts = counts
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
77 footerbar = "{:s}{:s} |{:s} |".format(self.str_tag, "-" * 12, "-" * 12,)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
78 if self.minmax is not None:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
79 ret.append(
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
80 "Trimmed to range (%s - %s)"
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
81 % (str(self.minmax[0]), str(self.minmax[1]))
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
82 )
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
83 for cutoff, original_count, scaled_count in zip(
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
84 cutoffs, counts, scaled_counts
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
85 ):
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
86 ret.append(
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
87 "{:s}{:>12.2f} |{:>12,d} | {:s}".format(
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
88 self.str_tag, cutoff, original_count, "*" * int(scaled_count)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
89 )
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
90 )
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
91 ret.append(footerbar)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
92 ret.append("{:s}{:>12s} |{:>12,d} |".format(self.str_tag, "N=", total))
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
93 ret.append(footerbar)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
94 ret.append('')
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
95 else:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
96 ret = []
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
97 if not self.generate_only:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
98 for line in ret:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
99 print(line)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
100 ret = "\n".join(ret)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
101 return ret
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
102
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
103
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
104 class findOut:
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
105
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
106 def __init__(self, args):
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
107 self.bwnames = args.bigwig
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
108 self.bwlabels = args.bigwiglabels
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
109 self.bedwin = args.minwin
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
110 self.outbeds = args.outbeds
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
111 self.bedouthi = args.bedouthi
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
112 self.bedoutlo = args.bedoutlo
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
113 self.bedouthilo = args.bedouthilo
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
114 self.tableoutfile = args.tableoutfile
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
115 self.bedwin = args.minwin
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
116 self.qlo = None
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
117 self.qhi = None
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
118 if args.outbeds != "outtab":
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
119 self.qhi = args.qhi
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
120 if args.qlo:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
121 try:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
122 f = float(args.qlo)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
123 self.qlo = f
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
124 except Exception:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
125 print('qlo not provided')
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
126 nbw = len(args.bigwig)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
127 nlab = len(args.bigwiglabels)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
128 if nlab < nbw:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
129 self.bwlabels += ["Nolabel"] * (nbw - nlab)
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
130 self.makeBed()
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
131
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
132 def processVals(self, bw, isTop):
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
133 """
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
134 idea from http://gregoryzynda.com/python/numpy/contiguous/interval/2019/11/29/contiguous-regions.html
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
135 Fast segmentation into regions by taking np.diff on the boolean array of over (under) cutpoint indicators in bwex.
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
136 This only gives non-zero values at the segment boundaries where there's a change, so those zeros are all removed in bwexdnz
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
137 leaving an array of segment start/end positions. That's twisted around into an array of start/end coordinates.
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
138 Magical. Fast. Could do the same for means or medians over windows for sparse bigwigs like repeat regions.
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
139 """
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
140 if isTop:
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
141 bwex = np.r_[False, bw >= self.bwtop, False] # extend with 0s
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
142 else:
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
143 bwex = np.r_[False, bw <= self.bwbot, False]
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
144 bwexd = np.diff(bwex)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
145 bwexdnz = bwexd.nonzero()[0]
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
146 bwregions = np.reshape(bwexdnz, (-1, 2))
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
147 return bwregions
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
148
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
149 def writeBed(self, bed, bedfname):
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
150 """
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
151 potentially multiple
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
152 """
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
153 bed.sort()
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
154 beds = ["%s\t%d\t%d\t%s\t%d" % x for x in bed]
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
155 with open(bedfname, "w") as bedf:
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
156 bedf.write("\n".join(beds))
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
157 bedf.write("\n")
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
158
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
159 def makeTableRow(self, bw, bwlabel, chr):
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
160 """
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
161 called for every contig, but messy inline
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
162 """
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
163 bwmean = np.mean(bw)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
164 bwstd = np.std(bw)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
165 bwmax = np.max(bw)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
166 nrow = np.size(bw)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
167 bwmin = np.min(bw)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
168 row = "%s\t%s\t%d\t%f\t%f\t%f\t%f" % (
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
169 bwlabel,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
170 chr,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
171 nrow,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
172 bwmean,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
173 bwstd,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
174 bwmin,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
175 bwmax,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
176 )
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
177 if self.qhi is not None:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
178 row += "\t%.2f" % self.bwtop
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
179 else:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
180 row += "\tnoqhi"
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
181 if self.qlo is not None:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
182 row += "\t%.2f" % self.bwbot
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
183 else:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
184 row += "\tnoqlo"
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
185 return row
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
186
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
187 def makeBed(self):
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
188 bedhi = []
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
189 bedlo = []
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
190 restab = []
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
191 bwlabels = self.bwlabels
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
192 bwnames = self.bwnames
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
193 reshead = "bigwig\tcontig\tn\tmean\tstd\tmin\tmax\tqtop\tqbot"
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
194 for i, bwname in enumerate(bwnames):
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
195 bwlabel = bwlabels[i].replace(" ", "")
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
196 fakepath = "in%d.bw" % i
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
197 if os.path.isfile(fakepath):
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
198 os.remove(fakepath)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
199 p = Path(fakepath)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
200 p.symlink_to(bwname) # required by pybigtools (!)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
201 bwf = pybigtools.open(fakepath)
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
202 chrlist = bwf.chroms()
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
203 chrs = list(chrlist.keys())
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
204 for chr in chrs:
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
205 first_few = None
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
206 bw = bwf.values(chr)
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
207 values, counts = np.unique(bw, return_counts=True)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
208 nvalues = len(values)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
209 if nvalues <= 20:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
210 histo = '\n'.join(['%s: %f occurs %d times' % (chr, values[x], counts[x]) for x in range(len(values))])
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
211 else:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
212 last10 = range(nvalues-10, nvalues)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
213 first_few = ['%.2f\t%d' % (values[x],counts[x]) for x in range(10)]
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
214 first_few += ['%.2f\t%d' % (values[x],counts[x]) for x in last10]
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
215 first_few.insert(0,'First/Last 10 value counts\nValue\tCount')
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
216 ha = asciihist(data=bw, bins=20, str_tag=chr)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
217 histo = ha.draw()
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
218 histo = '\n'.join(first_few) + '\nHistogram of bigwig values\n' + histo
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
219 bw = bw[~np.isnan(bw)] # some have NaN if parts of a contig not covered
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
220 if self.qhi is not None:
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
221 self.bwtop = np.quantile(bw, self.qhi)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
222 bwhi = self.processVals(bw, isTop=True)
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
223 for j, seg in enumerate(bwhi):
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
224 seglen = seg[1] - seg[0]
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
225 if seglen >= self.bedwin:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
226 score = np.sum(bw[seg[0]:seg[1]])/float(seglen)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
227 bedhi.append(
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
228 (
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
229 chr,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
230 seg[0],
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
231 seg[1],
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
232 "%s_%d" % (bwlabel, score),
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
233 score,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
234 )
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
235 )
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
236 if self.qlo is not None:
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
237 self.bwbot = np.quantile(bw, self.qlo)
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
238 bwlo = self.processVals(bw, isTop=False)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
239 for j, seg in enumerate(bwlo):
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
240 if seg[1] - seg[0] >= self.bedwin:
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
241 score = -1 * np.sum(bw[seg[0]:seg[1]])/float(seglen)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
242 bedlo.append(
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
243 (
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
244 chr,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
245 seg[0],
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
246 seg[1],
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
247 "%s_%d" % (bwlabel, score),
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
248 score,
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
249 )
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
250 )
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
251 if self.tableoutfile:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
252 row = self.makeTableRow(bw, bwlabel, chr)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
253 resheadl = reshead.split('\t')
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
254 rowl = row.split()
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
255 desc = ['%s\t%s' % (resheadl[x], rowl[x]) for x in range(len(rowl))]
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
256 desc.insert(0, 'Descriptive measures')
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
257 descn = '\n'.join(desc)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
258 restab.append(descn)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
259 restab.append(histo)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
260 if os.path.isfile(fakepath):
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
261 os.remove(fakepath)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
262 if self.tableoutfile:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
263 stable = "\n".join(restab)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
264 with open(self.tableoutfile, "w") as t:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
265 t.write(stable)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
266 t.write("\n")
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
267 some = False
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
268 if self.qlo:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
269 if self.outbeds in ["outall", "outlo", "outlohi"]:
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
270 self.writeBed(bedlo, self.bedoutlo)
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
271 some = True
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
272 if self.qhi:
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
273 if self.outbeds in ["outall", "outlohi", "outhi"]:
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
274 self.writeBed(bedhi, self.bedouthi)
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
275 some = True
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
276 if self.outbeds in ["outall", "outhilo"]:
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
277 allbed = bedlo + bedhi
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
278 self.writeBed(allbed, self.bedouthilo)
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
279 some = True
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
280 if not ((self.outbeds == 'outtab') or some):
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
281 sys.stderr.write(
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
282 "Invalid configuration - no output could be created. Was qlo missing and only low output requested for example?"
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
283 )
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
284 sys.exit(2)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
285
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
286
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
287 if __name__ == "__main__":
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
288 parser = argparse.ArgumentParser()
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
289 a = parser.add_argument
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
290 a("-m", "--minwin", default=10, type=int)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
291 a("-l", "--qlo", default=None)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
292 a("-i", "--qhi", default=None, type=float)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
293 a("--bedouthi", default=None)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
294 a("--bedoutlo", default=None)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
295 a("--bedouthilo", default=None)
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
296 a("-w", "--bigwig", nargs="+")
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
297 a("-n", "--bigwiglabels", nargs="+")
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
298 a("-o", "--outbeds", default="outhilo", help="optional high and low combined bed")
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
299 a("-t", "--tableoutfile", default=None)
0
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
300 args = parser.parse_args()
6
eb17eb8a3658 planemo upload commit 1baff96e75def9248afdcf21edec9bdc7ed42b1f-dirty
fubar
parents: 0
diff changeset
301 findOut(args)