benchmark_vcf_io.py

from pysam import VariantFile
import time
def count_oldvar_reg(vcffile_reg, labels=["OLDVARS", "OLDVARS_A", "OLDVARS_B"]):
    t_start = time.time()
    
    bcf_in = VariantFile(vcffile_reg)
    total = 0
    reg = 0
    
    for rec in bcf_in:
        total += 1
        for label in labels:
            if label in rec.info:
                reg += 1
                break
    print "Total: {0}".format(total)
    print "Found: {0}".format(reg)
    t_stop = time.time()
    print "Time takes: {0} s.".format(t_stop - t_start)

from vcf import Reader
def count_oldvar(vcffile_reg):
    t_start = time.time()
    
    reader = Reader(filename=vcffile_reg)
    total = 0
    reg = 0
    
    for variant in reader:
        total += 1
        for label in ["OLDVARS", "OLDVARS_A", "OLDVARS_B"]:
            if label in variant.INFO:
                reg += 1
                break
    print "Total: {0}".format(total)
    print "Reg: {0}".format(reg)
    t_stop = time.time()
    print "Time takes: {0} s.".format(t_stop - t_start)