Mercurial > repos > iuc > irma

annotate createMissingFiles.py @ 3:205a59cb55f1 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 308791d86112ad5ecfca99789841a9520e9bcb34
author iuc
date Wed, 29 Jan 2025 08:19:15 +0000
parents 736090e99c59
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
1 import glob
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
2 import os
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
3 import subprocess
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
4
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
5 dirPrefix = "resultDir/"
1
736090e99c59 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 6b8463ba27d0c91b736d579b0891632d4c032402
iuc
parents: 0
diff changeset
6 expectedSegments = {"A_MP": 7, "A_NP": 5, "A_HA": 4, "A_PB1": 2,
736090e99c59 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 6b8463ba27d0c91b736d579b0891632d4c032402
iuc
parents: 0
diff changeset
7 "A_PB2": 1, "A_NA": 6, "A_PA": 3, "A_NS": 8}
0
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
8
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
9
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
10 def renameSubtypeFiles(identifier):
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
11 files = glob.glob(dirPrefix + "A_" + identifier + "_*.*")
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
12 for file in files:
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
13 ext = file.split('.')[-1]
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
14 os.rename(file, dirPrefix + "A_" + identifier + "." + ext)
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
15
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
16
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
17 def getMissingSegments():
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
18 presentSegments = []
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
19 for file in os.listdir(dirPrefix):
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
20 if file.endswith(".fasta"):
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
21 presentSegments.append(file.split('.')[0])
1
736090e99c59 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 6b8463ba27d0c91b736d579b0891632d4c032402
iuc
parents: 0
diff changeset
22 return [segment for segment in expectedSegments.keys()
0
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
23 if segment not in presentSegments]
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
24
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
25
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
26 def getBamHeaderFromAnyFile():
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
27 anyBamFile = glob.glob(dirPrefix + "*.bam")[0]
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
28 samtoolsCmd = ["samtools", "view", "-H", anyBamFile]
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
29 result = subprocess.check_output(samtoolsCmd, text=True)
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
30 return result.split('\n')[0]
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
31
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
32
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
33 def getVcfHeaderFromAnyFile():
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
34 with open(glob.glob(dirPrefix + "*.vcf")[0]) as f:
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
35 anyVersionAndDateLines = f.readline() + f.readline()
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
36 emptyHeaderLine = "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO"
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
37 return anyVersionAndDateLines + emptyHeaderLine
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
38
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
39
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
40 def writeEmptyBam(identifier, bamHeader):
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
41 with open("headerSamFile.sam", "w") as f:
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
42 f.write(bamHeader) # write header to a temporary sam file
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
43 cmd = ['samtools', 'view', '-H', '-b', 'headerSamFile.sam']
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
44 targetBam = dirPrefix + identifier + ".bam"
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
45 with open(targetBam, "xb") as tB:
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
46 subprocess.check_call(cmd, stdout=tB)
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
47 os.remove("headerSamFile.sam")
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
48
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
49
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
50 def writeEmptyFasta(identifier):
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
51 open(dirPrefix + identifier + ".fasta", 'x').close()
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
52
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
53
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
54 def writeEmptyVcf(identifier, vcfHeader):
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
55 with open(dirPrefix + identifier + ".vcf", 'x') as f:
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
56 f.write(vcfHeader)
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
57
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
58
1
736090e99c59 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 6b8463ba27d0c91b736d579b0891632d4c032402
iuc
parents: 0
diff changeset
59 def writeEmptyAmendedFasta(identifier):
736090e99c59 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 6b8463ba27d0c91b736d579b0891632d4c032402
iuc
parents: 0
diff changeset
60 # irma names these files like: resultDir/amended_consensus/resultDir_<segNr>.fa
736090e99c59 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 6b8463ba27d0c91b736d579b0891632d4c032402
iuc
parents: 0
diff changeset
61 open(dirPrefix + "amended_consensus/resultDir_" + str(expectedSegments[identifier]) + ".fa", 'x').close()
736090e99c59 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 6b8463ba27d0c91b736d579b0891632d4c032402
iuc
parents: 0
diff changeset
62
736090e99c59 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 6b8463ba27d0c91b736d579b0891632d4c032402
iuc
parents: 0
diff changeset
63
0
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
64 def samtoolsSortAllBam():
1
736090e99c59 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 6b8463ba27d0c91b736d579b0891632d4c032402
iuc
parents: 0
diff changeset
65 for segment in expectedSegments.keys():
0
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
66 os.rename(dirPrefix + segment + ".bam",
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
67 dirPrefix + segment + "_unsorted.bam")
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
68 cmd = ['samtools', 'sort', dirPrefix + segment + "_unsorted.bam"]
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
69 targetBam = dirPrefix + segment + ".bam"
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
70 with open(targetBam, "w") as tB:
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
71 subprocess.check_call(cmd, stdout=tB, text=True)
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
72 os.remove(dirPrefix + segment + "_unsorted.bam")
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
73
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
74
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
75 if __name__ == "__main__":
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
76 renameSubtypeFiles("HA")
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
77 renameSubtypeFiles("NA")
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
78 bamHeader = getBamHeaderFromAnyFile()
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
79 vcfHeader = getVcfHeaderFromAnyFile()
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
80 for segment in getMissingSegments():
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
81 writeEmptyBam(segment, bamHeader)
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
82 writeEmptyFasta(segment)
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
83 writeEmptyVcf(segment, vcfHeader)
1
736090e99c59 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 6b8463ba27d0c91b736d579b0891632d4c032402
iuc
parents: 0
diff changeset
84 writeEmptyAmendedFasta(segment)
0
3d86c05cd838 planemo upload for repository https://github.com/aaronKol/tools-iuc/tree/main/tools/irma commit 0ee665c3393af083833fdb9becbe6965d009e16c
iuc
parents:
diff changeset
85 samtoolsSortAllBam()