generate sample_info according info.txt, pn and disease

get_sample_info.py

#!/usr/bin/env python
#-*- coding: utf-8 -*-

 
def get_sample_info(infofile, pn, disease):
    header = '#B1\n#FamilyID\tSampleID\tSex\tNormal/Patient\tPN'
    if disease:
        header += '\tDisease'
    header += '\n'

    with open('sample_info', 'w') as out:
        out.write(header)
        with open(infofile) as f:
            for eachline in f:
                if len(eachline) > 1:                
                    familyid = eachline.rstrip().split('\t')[0] or '.'
                    sampleid = eachline.rstrip().split('\t')[2]
                    sex = eachline.rstrip().split('\t')[4]
                    nop = eachline.rstrip().split('\t')[6]
                    line = '{}\t{}\t{}\t{}\t{}'.format(familyid, sampleid, sex, nop, pn)
                    if disease:
                        line = line + '\t' + disease
                    line += '\n'
                    out.write(line)
            

if __name__ == '__main__':
    import sys

    if len(sys.argv) < 3:
        print "Usage: {} <info.txt> <pn> [disease]".format(sys.argv[0])
        exit(1)
        
    if len(sys.argv) == 4:
        disease = sys.argv[3]
    else:
        disease = None

    get_sample_info(sys.argv[1], sys.argv[2], disease)