Mercurial > repos > cstrittmatter > test_eurl_vtec_wgs_pt

annotate scripts/ReMatCh/utils/gffParser.py @ 0:965517909457 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
author cstrittmatter
date Wed, 22 Jan 2020 08:41:44 -0500
parents
children 0cbed1c0a762
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
1 #!/usr/bin/env python
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
2
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
3 import argparse, sys, os
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
4 from Bio import SeqIO
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
5 from Bio.Seq import Seq
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
6 from Bio.SeqRecord import SeqRecord
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
7 from Bio.Alphabet import IUPAC
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
8 import ntpath
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
9
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
10 def parseID(filename):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
11 #get wanted feature IDs
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
12 gffIDs=[]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
13 with open(filename, 'r') as in_handle:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
14 for line in in_handle:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
15 line=line.strip()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
16 gffIDs.append(line)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
17 return gffIDs
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
18
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
19 def retrieveSeq_File(fastaFile, coordFile, extraSeq, filename, outputDir):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
20 #Parsing the sequence file, using the provided txt file containing the contig ID and positions to retrieve sequences.
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
21 handle = open(fastaFile, "rU")
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
22 records_dict = SeqIO.to_dict(SeqIO.parse(handle, "fasta"))
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
23 handle.close()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
24
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
25 Seq2Get={}
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
26 with open(coordFile, 'r') as sequeces2get:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
27 for line in sequeces2get:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
28 line=line.split(',')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
29 coords=(int(line[-2]), int(line[-1]))
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
30 contigID=line[0]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
31 if contigID in Seq2Get.keys():
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
32 Seq2Get[contigID].append(coords)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
33 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
34 Seq2Get[contigID]=[coords]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
35
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
36 with open(outputDir+'/'+filename+'.fasta','w') as output_handle:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
37 fails=0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
38 successes=0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
39 records=[]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
40 for contig, listCoords in Seq2Get.items():
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
41 contigSeq=records_dict[contig].seq
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
42 for coord in listCoords:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
43 coord1=coord[0]-extraSeq
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
44 coord2=coord[1]+extraSeq
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
45 if coord1 < 0 or coord2 > len(contigSeq):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
46 fail_log=open(outputDir+'/'+filename+'_fails.txt', 'a')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
47 fail_log.write(contig + ',' + str(coord[0])+','+ str(coord[1])+'\n')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
48 fail_log.close()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
49 fails+=1
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
50 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
51 geneseq=str(contigSeq[coord1:coord2])
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
52 record = SeqRecord(Seq(geneseq), id=str(str(contig)+'#'+str(coord1)+'_'+str(coord2)), description='')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
53 records.append(record)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
54 successes+=1
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
55 SeqIO.write(records, output_handle, "fasta")
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
56
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
57 print 'Retrived %s features successfully from %s with %s bp as extra sequence.' % (str(successes), filename, str(extraSeq))
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
58 if fails>0:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
59 print '%s featrued failed to retrieve. Check %s_fails.txt file.' % (str(fails), filename)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
60
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
61 def retrieveSeq(fastaFile, gffFeatures, extraSeq, filename, outputDir):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
62 #parsing the sequence file into a SeqIO dictionary. one contig per entry
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
63 handle = open(fastaFile, "rU")
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
64 records_dict = SeqIO.to_dict(SeqIO.parse(handle, "fasta"))
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
65 handle.close()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
66
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
67 with open(outputDir+'/'+filename+'.fasta','w') as output_handle:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
68 fails=0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
69 successes=0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
70 records=[]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
71 for locus, location in gffFeatures.items():
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
72 #print locus
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
73 contigSeq=records_dict[location[0]].seq
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
74 coord1=location[1]-extraSeq
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
75 coord2=location[2]+extraSeq
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
76 if coord1 < 0 or coord2 > len(contigSeq):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
77 fail_log=open(outputDir+'/'+filename+'_fails.txt', 'a')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
78 fail_log.write(locus+'\n')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
79 fail_log.close()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
80 fails+=1
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
81 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
82 geneseq=str(contigSeq[coord1:coord2])
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
83 if location[3] == '-':
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
84 seq = Seq(geneseq)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
85 geneseq =str(seq.reverse_complement())
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
86 record = SeqRecord(Seq(geneseq), id=str(locus+'-'+str(location[0])+'#'+str(location[1])+'_'+str(location[2])), description='')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
87 records.append(record)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
88 successes+=1
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
89 SeqIO.write(records, output_handle, "fasta")
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
90 print 'Retrived %s features successfully from %s with %s bp as extra sequence.' % (str(successes), filename, str(extraSeq))
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
91 if fails>0:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
92 print '%s featrued failed to retrieve. Check %s_fails.txt file.' % (str(fails), filename)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
93
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
94 def parseFeatures(tempGFF):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
95 #parsing the feature file into a dictionary
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
96 gffFeatures={}
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
97
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
98 with open(tempGFF, 'r') as temp_genes:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
99 for line in temp_genes:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
100 line=line.split('\t')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
101 if "CDS" in line[2]:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
102 ID=line[-1].split(';')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
103 locusID=str(ID[0].split('=')[1])
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
104 contig=line[0]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
105 begining=int(line[3])-1 #to get the full sequence
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
106 end=int(line[4])
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
107 strand=line[6]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
108 location=[contig, begining, end, strand]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
109 gffFeatures[locusID]=location
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
110 return gffFeatures
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
111
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
112 def gffParser(gffFile, extraSeq=0, outputDir='.', keepTemporaryFiles=False, IDs=None, coordFile=None):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
113
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
114 filename=ntpath.basename(gffFile).replace('.gff', '')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
115
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
116 #cleaning temp files if they exist
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
117 if os.path.isfile(outputDir+'/'+filename+'_features.gff'):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
118 os.remove(outputDir+'/'+filename+'_features.gff')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
119 if os.path.isfile(outputDir+'/'+filename+'_sequence.fasta'):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
120 os.remove(outputDir+'/'+filename+'_sequence.fasta')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
121
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
122 #cleaning fails file if it exists
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
123 if os.path.isfile(outputDir+'/'+filename+'_fails.txt'):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
124 os.remove(outputDir+'/'+filename+'_fails.txt')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
125
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
126 if coordFile is None:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
127
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
128 if IDs is not None:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
129 selectIDs=parseID(IDs)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
130 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
131 selectIDs=None
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
132
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
133 #separating the gff into 2 different files: one with the features and another with the conting sequences
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
134 with open(gffFile, 'r') as in_handle, open(outputDir+'/'+filename+'_features.gff', 'a') as temp_genes, open(outputDir+'/'+filename+'_sequence.fasta', 'a') as temp_contigs:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
135 for line in in_handle:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
136 if not line.startswith('##'):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
137 if '\t' in line:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
138 if selectIDs is not None:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
139 items=line.split('\t')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
140 ID=items[-1].split(';')[0]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
141 ID=ID.split('=')[1]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
142 if ID in selectIDs:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
143 temp_genes.write(line)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
144 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
145 temp_genes.write(line)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
146 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
147 temp_contigs.write(line)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
148
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
149 gffFiles=parseFeatures(outputDir+'/'+filename+'_features.gff')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
150
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
151 retrieveSeq(outputDir+'/'+filename+'_sequence.fasta', gffFiles, extraSeq, filename, outputDir)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
152
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
153 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
154 with open(gffFile, 'r') as in_handle, open(outputDir+'/'+filename+'_sequence.fasta', 'a') as temp_contigs:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
155 for line in in_handle:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
156 if not line.startswith('##'):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
157 if '\t' in line:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
158 pass
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
159 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
160 temp_contigs.write(line)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
161
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
162 retrieveSeq_File(outputDir+'/'+filename+'_sequence.fasta', coordFile, extraSeq, filename, outputDir)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
163
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
164 #removing temp files
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
165 if not keepTemporaryFiles:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
166 try:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
167 os.remove(outputDir+'/'+filename+'_features.gff')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
168 except:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
169 pass
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
170 os.remove(outputDir+'/'+filename+'_sequence.fasta')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
171
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
172 def main():
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
173
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
174 version='1.0.0'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
175
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
176 parser = argparse.ArgumentParser(prog='gffParser.py', description='GFF3 parser for feature sequence retrival.', epilog='by C I Mendes (cimendes@medicina.ulisboa.pt)')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
177 parser.add_argument('--version', help='Version information', action='version', version=str('%(prog)s v' + version))
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
178
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
179 parser.add_argument('-i', '--input', help='GFF3 file to parse, containing both sequences and annotations (like the one obtained from PROKKA).', type=argparse.FileType('r'), required=True)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
180 parser.add_argument('-x', '--extraSeq', help='Extra sequence to retrieve per feature in gff.', default=0, type=int, required=False)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
181 parser.add_argument('-k','--keepTemporaryFiles', help='Keep temporary gff(without sequence) and fasta files.', action='store_true')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
182 parser.add_argument('-o', '--outputDir', help='Path to where the output is to be saved.', default='.', required=False)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
183
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
184
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
185 parser_optional_selected_regions_exclusive = parser.add_mutually_exclusive_group()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
186 parser_optional_selected_regions_exclusive.add_argument('-s', '--select', help='txt file with the IDs of interest, one per line', type=argparse.FileType('r'), required=False)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
187 parser_optional_selected_regions_exclusive.add_argument('-f', '--fromFile', help='Sequence coordinates to be retrieved. Requires contig ID and coords (contig,strart,end) in a csv file. One per line.', type=argparse.FileType('r'), required=False)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
188
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
189 args = parser.parse_args()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
190
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
191 gffParser(os.path.abspath(args.input.name), args.extraSeq, os.path.abspath(args.outputDir), args.keepTemporaryFiles, os.path.abspath(args.select.name) if args.select is not None else None, os.path.abspath(args.fromFile.name) if args.fromFile is not None else None)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
192
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
193
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
194 if __name__ == "__main__":
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
195 main()