Mercurial > repos > drosofff > blastn_to_scaffold
annotate blastn_to_scaffold.py @ 0:7b4212122a49 draft
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
author | drosofff |
---|---|
date | Fri, 12 Jun 2015 12:27:30 -0400 |
parents | |
children | ef007adcbbf8 |
rev | line source |
---|---|
0
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
1 #!/usr/bin/python |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
2 import sys |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
3 import argparse |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
4 |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
5 def insert_newlines(string, every=60): |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
6 lines = [] |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
7 for i in xrange(0, len(string), every): |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
8 lines.append(string[i:i+every]) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
9 return '\n'.join(lines) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
10 |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
11 def getseq (fastadict, transcript, up, down, orientation="direct"): |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
12 def reverse (seq): |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
13 revdict = {"A":"T","T":"A","G":"C","C":"G","N":"N"} |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
14 revseq = [revdict[i] for i in seq[::-1]] |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
15 return "".join(revseq) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
16 pickseq = fastadict[transcript][up-1:down] |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
17 if orientation == "direct": |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
18 return pickseq |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
19 else: |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
20 return reverse(pickseq) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
21 |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
22 def Parser(): |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
23 the_parser = argparse.ArgumentParser( |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
24 description="Generate DNA scaffold from blastn alignment of Contigs") |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
25 the_parser.add_argument( |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
26 '--sequences', action="store", type=str, help="input sequence file in fasta format") |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
27 the_parser.add_argument( |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
28 '--guideSequence', action="store", type=str, help="the reference sequence to guide the scaffold assembly in fasta format") |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
29 the_parser.add_argument( |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
30 '--blastn-tab', dest="blastn_tab", action="store", type=str, help="13-columns tabular blastn output") |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
31 the_parser.add_argument( |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
32 '--output', action="store", type=str, help="output file path, fasta format") |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
33 args = the_parser.parse_args() |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
34 return args |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
35 |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
36 def blatnInfo (file): |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
37 blastlist = [] |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
38 with open(file, "r") as f: |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
39 for line in f: |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
40 minilist = [] |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
41 fields = line.rstrip().split() |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
42 minilist.append(fields[0]) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
43 minilist.extend(fields[6:10]) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
44 blastlist.append(minilist) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
45 blastlist.sort(key=lambda x: x[3], reverse=True) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
46 return blastlist |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
47 |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
48 def myContigs (file): |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
49 Contigs = {} |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
50 with open(file, "r") as f: |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
51 for line in f: |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
52 if line[0] == ">": |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
53 header = line[1:-1] |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
54 Contigs[header] = "" |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
55 else: |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
56 Contigs[header] += line[:-1] |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
57 return Contigs |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
58 |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
59 def myGuide (file): |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
60 Guide = {} |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
61 coordinate = 0 |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
62 with open(file, "r") as f: |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
63 for line in f: |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
64 if line[0] == ">": |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
65 continue |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
66 else: |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
67 for nucleotide in line[:-1]: |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
68 coordinate += 1 |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
69 Guide[coordinate] = nucleotide.lower() |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
70 return Guide |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
71 |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
72 def updateGuide (blastlist, GuideDict, ContigsDict): |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
73 ''' |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
74 the blastlist object is a list of list with |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
75 element [0] : name of the blasted Contig |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
76 element [1] : queryStart of the alignment to the reference |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
77 element [2] = queryStop of the alignment to the reference |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
78 element [3] : subjectStart of the alignment to the reference |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
79 element [4] = subjectStop of the alignment to the reference |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
80 ''' |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
81 for fields in blastlist: |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
82 seqHeader = fields[0] |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
83 queryStart = int(fields[1]) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
84 queryStop = int(fields[2]) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
85 subjectStart = int(fields[3]) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
86 subjectStop = int(fields[4]) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
87 if subjectStart > subjectStop: |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
88 subjectStart, subjectStop = subjectStop, subjectStart |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
89 orientation = "reverse" |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
90 else: |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
91 orientation = "direct" |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
92 sequence = getseq (ContigsDict, seqHeader, queryStart, queryStop, orientation) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
93 for i in range(subjectStart, subjectStop+1): |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
94 del GuideDict[i] |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
95 for i, nucleotide in enumerate(sequence): |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
96 GuideDict[i+subjectStart] = nucleotide |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
97 |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
98 def finalAssembly (GuideDict, outputfile): |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
99 finalSeqList = [] |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
100 for keys in sorted(GuideDict): |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
101 finalSeqList.append(GuideDict[keys]) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
102 finalSequence = insert_newlines("".join(finalSeqList) ) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
103 Out = open (outputfile, "w") |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
104 print >> Out, ">Scaffold" |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
105 print >> Out, finalSequence |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
106 Out.close() |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
107 |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
108 def __main__(): |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
109 args = Parser() |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
110 ContigsDict = myContigs (args.sequences) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
111 GuideDict = myGuide (args.guideSequence) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
112 blastlist = blatnInfo(args.blastn_tab) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
113 updateGuide(blastlist, GuideDict, ContigsDict) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
114 finalAssembly(GuideDict, args.output) |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
115 |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
116 if __name__ == "__main__": |
7b4212122a49
planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff
changeset
|
117 __main__() |