Mercurial > repos > fubar > bigwig_outlier_bed

annotate bigwig_outlier_bed_slow1.py @ 1:04129d5129be draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
author fubar
date Mon, 01 Jul 2024 02:57:16 +0000
parents c71db540eb38
children
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 import os
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
2 import pyBigWig
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
3
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
4
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
5 class findOut():
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
6
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
7 def __init__(self, bwname="test.bw", bedname="test.bed", sd_lo=3, sd_hi=3, bedwin=10):
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
8 self.bedwin = bedwin
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
9 self.bwf = pyBigWig.open(bwname, "r")
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
10 self.chrlist = self.bwf.chroms()
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
11 self.bedf = open(bedname, "w")
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
12 self.sd_lo = sd_lo
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
13 self.sd_hi = sd_hi
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
14 self.makeBed(self.chrlist)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
15
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
16 def makeBed(self, chrlist):
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
17 bed = []
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
18 for chr in chrlist:
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
19 print(chr)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
20 gm = self.bwf.stats(chr, 0, type="mean")[0]
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
21 gstd = self.bwf.stats(chr, 0, type="std")[0]
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
22 cutlow = gm - (self.sd_lo * gstd)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
23 cuthi = gm + (self.sd_hi * gstd)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
24 chr_len = self.chrlist[chr]
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
25 nb = chr_len // self.bedwin
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
26 means = self.bwf.stats(chr, 0, chr_len, type="mean", nBins=nb)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
27 inlo = False
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
28 inhi = False
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
29 reg_start = None
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
30 reg_end = None
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
31 reg_means = []
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
32 print('got %d means, gm=%f, lo=%f, hi=%f' % (len(means), gm, cutlow, cuthi))
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
33 for i, m in enumerate(means):
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
34 bend = min(chr_len, (i + 1) * self.bedwin)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
35 featname = "%s_%d" % (chr,i)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
36 if m and (m < cutlow or m > cuthi):
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
37 if inlo:
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
38 if m < cutlow: # extend
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
39 reg_end = bend
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
40 reg_means.append(m)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
41 else: # high so close
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
42 rm = sum(reg_means) / len(reg_means)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
43 bed.append(
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
44 "%s\t%d\t%d\t%s\t%.3f\n"
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
45 % (chr, reg_start, reg_end, featname, rm)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
46 )
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
47 inlo = False
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
48 reg_means = []
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
49 elif inhi:
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
50 if m > cuthi: # extend
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
51 reg_end = bend
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
52 reg_means.append(m)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
53 else:
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
54 rm = sum(reg_means) / len(reg_means)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
55 bed.append(
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
56 "%s\t%d\t%d\t%s\t%.3f\n"
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
57 % (chr, reg_start, reg_end, featname, rm)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
58 )
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
59 inhi = False
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
60 reg_means = []
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
61 elif m < cutlow: # start new low region
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
62 inlo = True
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
63 reg_start = i * self.bedwin
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
64 reg_end = bend
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
65 reg_means = [m]
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
66 elif m > cuthi:
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
67 inhi = True
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
68 reg_start = i * self.bedwin
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
69 reg_end = bend
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
70 reg_means = [m]
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
71 else: # not out of range - write current extended bed region
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
72 if inhi or inlo:
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
73 inhi = False
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
74 inlo = False
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
75 rm = sum(reg_means) / len(reg_means)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
76 bed.append(
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
77 "%s\t%d\t%d\t%s\t%.3f\n"
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
78 % (chr, reg_start, reg_end, featname, rm)
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
79 )
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
80 reg_means = []
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
81 reg_start = None
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
82 reg_end = None
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
83 self.bedf.write(''.join(bed))
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
84 self.bedf.close()
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
85
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
86
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
87 if __name__ == "__main__":
c71db540eb38 planemo upload for repository https://github.com/jackh726/bigtools commit ce6b9f638ebcebcad5a5b10219f252962f30e5cc-dirty
fubar
parents:
diff changeset
88 fo = findOut(bwname="test.bw", bedname="test.bed", sd_lo=2, sd_hi=2, bedwin=100)