comparison filter_fasta.py @ 5:af454e5a9ef5 draft

Uploaded
author bornea
date Sat, 06 Aug 2016 17:00:41 -0400
parents b25b0f210e10
children 5990c4dbfbaa
comparison
equal deleted inserted replaced
4:489b560129c8 5:af454e5a9ef5
1 # -*- coding: utf-8 -*- 1
2 """ 2 """
3 Python-code: Merge Scaffold Samples Report files 3 Python-code: Merge Scaffold Samples Report files
4 @author = Brent Kuenzi 4 @author = Brent Kuenzi
5 @email = Brent.Kuenzi@moffitt.org 5 @email = Brent.Kuenzi@moffitt.org
6 """ 6 """
20 output = [] 20 output = []
21 for line in x: 21 for line in x:
22 line = line.strip() 22 line = line.strip()
23 temp = line.split('\t') 23 temp = line.split('\t')
24 output.append(temp) 24 output.append(temp)
25 print temp
25 return output 26 return output
26 def getAccessions(infile): # get list of protein accessions from your data 27 def getAccessions(infile): # get list of protein accessions from your data
27 data = readtab(infile) 28 data = readtab(infile)
28 cnt = 0 29 cnt = 0
29 header_start = 0 30 header_start = 0
80 seq.append(merged) 81 seq.append(merged)
81 cnt=0 82 cnt=0
82 with open("output.fasta","wt") as x: 83 with open("output.fasta","wt") as x:
83 for i in header: 84 for i in header:
84 x.write(i+'\n'+seq[cnt]+'\n') 85 x.write(i+'\n'+seq[cnt]+'\n')
86 print i+'\n'+seq[cnt]+'\n'
85 cnt+=1 87 cnt+=1
86 fasta = sys.argv[1] # fasta file to filter 88 fasta = sys.argv[1] # fasta file to filter
87 data = sys.argv[2] # scaffold report #2 -- filename 89 data = sys.argv[2] # scaffold report #2 -- filename
88 90
89 FilterFastaSeq(fasta,getAccessions(data)) 91 FilterFastaSeq(fasta,getAccessions(data))