vcf-select

#!/usr/bin/env python
# -*- coding: utf-8 -*-
#### t.py
#### made by Daniel Minseok Kwon
#### 2020-08-31 12:50:13
#########################
import sys
import os
SVRNAME = os.uname()[1]
if "MBI" in SVRNAME.upper():
    sys_path="/Users/pcaso/bin/python_lib"
elif SVRNAME == "T7":
    sys_path="/ms1/bin/python_lib"
else:
    sys_path="/home/mk446/bin/python_lib"
sys.path.append(sys_path)

import file_util
import proc_util


def vcf_select(base_vcf, pos_target_vcf):
    out = base_vcf + '_' + pos_target_vcf + '.taget.vcf'
    m = {}
    for line in file_util.gzopen(pos_target_vcf):
        
        line = file_util.decodeb(line)
        if line[0] != '#':
            arr = line.split('\t')
            m[arr[0].replace('chr', '') + '_' + arr[1]]=1

    print('total target:', len(m.keys()))

    i = 0
    cnt = 0
    f = open(out, 'w')
    print('Saving..', out)
    for line in file_util.gzopen(base_vcf):
        line = file_util.decodeb(line)
        if line[0] == '#':
            f.write(line)
        else:
            arr = line.split('\t')
            try:
                m[arr[0].replace('chr', '') + '_' + arr[1]]
                # print(line, end="")
                f.write(line)
                cnt += 1
            except KeyError:
                pass
            if i % 1000000 == 0:
                print("\tprocessing.....",cnt, 'included.', i ,arr[:5])
            i += 1
    f.close()
    print('Saved..', out)


if __name__ == "__main__":
    print('vcf-select [BASE VCF] [POSITION TARGET VCF]')
    base_vcf = sys.argv[1]
    pos_target_vcf = sys.argv[2]
    vcf_select(base_vcf, pos_target_vcf)