annotate BlastParser_and_hits.py @ 0:3959a271cf3f draft

planemo upload for repository https://bitbucket.org/drosofff/gedtools/
author drosofff
date Tue, 09 Jun 2015 04:15:34 -0400
parents
children e0985bad7b92
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
1 #!/usr/bin/python
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
2 # blastn blastx parser revised debugged: 3-4-2015. Commit issue.
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
3 # drosofff@gmail.com
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
4
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
5 import sys
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
6 import argparse
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
7 from collections import defaultdict
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
8
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
9 def Parser():
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
10 the_parser = argparse.ArgumentParser()
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
11 the_parser.add_argument('--blast', action="store", type=str, help="Path to the blast output (tabular format, 12 column)")
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
12 the_parser.add_argument('--sequences', action="store", type=str, help="Path to the fasta file with blasted sequences")
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
13 the_parser.add_argument('--fastaOutput', action="store", type=str, help="fasta output file of blast hits")
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
14 the_parser.add_argument('--tabularOutput', action="store", type=str, help="tabular output file of blast analysis")
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
15 the_parser.add_argument('--flanking', action="store", type=int, help="number of flanking nucleotides added to the hit sequences")
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
16 args = the_parser.parse_args()
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
17 if not all ( (args.sequences, args.blast, args.fastaOutput, args.tabularOutput) ):
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
18 the_parser.error('argument(s) missing, call the -h option of the script')
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
19 if not args.flanking:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
20 args.flanking = 0
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
21 return args
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
22
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
23 def median(lst):
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
24 lst = sorted(lst)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
25 if len(lst) < 1:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
26 return None
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
27 if len(lst) %2 == 1:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
28 return lst[((len(lst)+1)/2)-1]
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
29 if len(lst) %2 == 0:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
30 return float(sum(lst[(len(lst)/2)-1:(len(lst)/2)+1]))/2.0
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
31
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
32 def getfasta (fastafile):
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
33 fastadic = {}
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
34 for line in open (fastafile):
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
35 if line[0] == ">":
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
36 header = line[1:-1]
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
37 fastadic[header] = ""
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
38 else:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
39 fastadic[header] += line
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
40 for header in fastadic:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
41 fastadic[header] = "".join(fastadic[header].split("\n"))
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
42 return fastadic
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
43
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
44 def insert_newlines(string, every=60):
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
45 lines = []
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
46 for i in xrange(0, len(string), every):
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
47 lines.append(string[i:i+every])
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
48 return '\n'.join(lines)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
49
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
50 def getblast (blastfile):
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
51 '''blastinfo [0] Percentage of identical matches
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
52 blastinfo [1] Alignment length
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
53 blastinfo [2] Number of mismatches
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
54 blastinfo [3] Number of gap openings
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
55 blastinfo [4] Start of alignment in query
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
56 blastinfo [5] End of alignment in query
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
57 blastinfo [6] Start of alignment in subject (database hit)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
58 blastinfo [7] End of alignment in subject (database hit)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
59 blastinfo [8] Expectation value (E-value)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
60 blastinfo [9] Bit score
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
61 blastinfo [10] Subject length (NEED TO BE SPECIFIED WHEN RUNNING BLAST) '''
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
62 blastdic = defaultdict (dict)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
63 for line in open (blastfile):
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
64 fields = line[:-1].split("\t")
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
65 transcript = fields[0]
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
66 subject = fields[1]
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
67 blastinfo = [float(fields[2]) ] # blastinfo[0]
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
68 blastinfo = blastinfo + [int(i) for i in fields[3:10] ] # blastinfo[1:8] insets 1 to 7
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
69 blastinfo.append(fields[10]) # blastinfo[8] E-value remains as a string type
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
70 blastinfo.append(float(fields[11])) # blastinfo[9] Bit score
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
71 blastinfo.append(int(fields[12])) # blastinfo[10] Subject length MUST BE RETRIEVED THROUGH A 13 COLUMN BLAST OUTPUT
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
72 try:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
73 blastdic[subject][transcript].append(blastinfo)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
74 except:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
75 blastdic[subject][transcript] = [ blastinfo ]
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
76 return blastdic
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
77
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
78 def getseq (fastadict, transcript, up, down, orientation="direct"):
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
79 def reverse (seq):
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
80 revdict = {"A":"T","T":"A","G":"C","C":"G","N":"N"}
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
81 revseq = [revdict[i] for i in seq[::-1]]
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
82 return "".join(revseq)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
83 pickseq = fastadict[transcript][up-1:down]
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
84 if orientation == "direct":
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
85 return pickseq
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
86 else:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
87 return reverse(pickseq)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
88
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
89 def subjectCoverage (fastadict, blastdict, subject, QueriesFlankingNucleotides=0):
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
90 SubjectCoverageList = []
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
91 HitDic = {}
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
92 bitScores = []
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
93 for transcript in blastdict[subject]:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
94 prefix = "%s--%s_" % (subject, transcript)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
95 hitNumber = 0
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
96 for hit in blastdict[subject][transcript]:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
97 hitNumber += 1
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
98 suffix = "hit%s_IdMatch=%s,AligLength=%s,E-val=%s" % (hitNumber, hit[0], hit[1], hit[8])
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
99 HitDic[prefix+suffix] = GetHitSequence (fastadict, transcript, hit[4], hit[5], QueriesFlankingNucleotides) #query coverage by a hit is in hit[4:6]
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
100 SubjectCoverageList += range (min([hit[6], hit[7]]), max([hit[6], hit[7]]) + 1) # subject coverage by a hit is in hit[6:8]
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
101 bitScores.append(hit[9])
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
102 subjectLength = hit [10] # always the same value for a given subject. Stupid but simple
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
103 TotalSubjectCoverage = len ( set (SubjectCoverageList) )
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
104 RelativeSubjectCoverage = TotalSubjectCoverage/float(subjectLength)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
105 return HitDic, subjectLength, TotalSubjectCoverage, RelativeSubjectCoverage, max(bitScores), median(bitScores)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
106
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
107 def GetHitSequence (fastadict, FastaHeader, leftCoordinate, rightCoordinate, FlankingValue):
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
108 if rightCoordinate > leftCoordinate:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
109 polarity = "direct"
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
110 else:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
111 polarity = "reverse"
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
112 leftCoordinate, rightCoordinate = rightCoordinate, leftCoordinate
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
113 if leftCoordinate - FlankingValue > 0:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
114 leftCoordinate -= FlankingValue
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
115 else:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
116 leftCoordinate = 1
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
117 return getseq (fastadict, FastaHeader, leftCoordinate, rightCoordinate, polarity)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
118
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
119 def __main__ ():
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
120 args = Parser()
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
121 fastadict = getfasta (args.sequences)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
122 Xblastdict = getblast (args.blast)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
123 results = defaultdict(dict)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
124 F = open(args.tabularOutput, "w")
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
125 Fasta = open(args.fastaOutput, "w")
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
126 for subject in Xblastdict:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
127 results[subject]["HitDic"], results[subject]["subjectLength"], results[subject]["TotalCoverage"], results[subject]["RelativeSubjectCoverage"], results[subject]["maxBitScores"], results[subject]["medianBitScores"] = subjectCoverage(fastadict, Xblastdict, subject, args.flanking)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
128 ## data output
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
129 print >>F, "# SeqId\t%Identity\tAlignLength\tStartSubject\tEndSubject\t%QueryHitCov\tE-value\tBitScore\n"
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
130 for subject in sorted (results, key=lambda x: results[x]["TotalCoverage"], reverse=True):
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
131 print >> F, "#\n# %s" % subject
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
132 print >> F, "# Suject Length: %s" % (results[subject]["subjectLength"])
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
133 print >> F, "# Total Subject Coverage: %s" % (results[subject]["TotalCoverage"])
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
134 print >> F, "# Relative Subject Coverage: %s" % (results[subject]["RelativeSubjectCoverage"])
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
135 print >> F, "# Maximum Bit Score: %s" % (results[subject]["maxBitScores"])
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
136 print >> F, "# Median Bit Score: %s" % (results[subject]["medianBitScores"])
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
137 for header in results[subject]["HitDic"]:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
138 print >> Fasta, ">%s\n%s" % (header, insert_newlines(results[subject]["HitDic"][header]) )
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
139 for transcript in Xblastdict[subject]:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
140 transcriptSize = float(len(fastadict[transcript]))
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
141 for hit in Xblastdict[subject][transcript]:
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
142 percentIdentity, alignLenght, subjectStart, subjectEnd, queryCov = hit[0], hit[1], hit[6], hit[7], "%.1f" % (abs(hit[5]-hit[4])/transcriptSize*100)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
143 Eval, BitScore = hit[8], hit[9]
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
144 info = [transcript] + [percentIdentity, alignLenght, subjectStart, subjectEnd, queryCov, Eval, BitScore]
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
145 info = [str(i) for i in info]
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
146 info = "\t".join(info)
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
147 print >> F, info
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
148 print >> Fasta, ""
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
149 F.close()
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
150 Fasta.close()
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
151
3959a271cf3f planemo upload for repository https://bitbucket.org/drosofff/gedtools/
drosofff
parents:
diff changeset
152 if __name__=="__main__": __main__()