cmp-bbi.py

from bx.bbi import bigwig_file
from bx.bbi import bigbed_file
print bigwig_file
import time
from hashlib import md5

prefix = '/data/gemini/data/hg19_fitcons_fc-i6-0_V1-01'


print "before bw: 8.6 seconds"
print "before bw: 8529aaef661eb479c2bfe6e1ee4107db"
print "before bb: 10.4 seconds"
print "before bb: d2424d770247e721b43c2729cbe160a0"


for suffix in ('bb', 'bw'):
    t = time.time()
    if suffix == 'bw': 
        bw = bigwig_file.BigWigFile(open("%s.%s" % (prefix, suffix)))
    else:
        bw = bigbed_file.BigBedFile(open("%s.%s" % (prefix, suffix)))

    for chrom in range(1, 20):
        bw.query('chr%i' % chrom, 100000, 200000, 20)
        bw.query('chr%i' % chrom, 100000, 200000, 2000)
        bw.query('chr%i' % chrom, 1000000, 1200000, 20)
        bw.query('chr%i' % chrom, 1000000, 1200000, 2000)
        bw.query('chr%i' % chrom, 1000000, 2000000, 20)
        bw.query('chr%i' % chrom, 1000000, 2000000, 2000)

        for i in range(100):
            bw.query('chr%i' % chrom, 1000001 + i, 1000002 + i, 1)

        if chrom == 19:
            vals = []
            for i in range(100):
                res = bw.query('chr%i' % chrom, 1000001 + i, 1000002 + i, 1)
                assert len(res) == 1
                res = res[0]
                val = "|".join("%s -> %.3f" % (k,v) for k, v in res.items())
                vals.append(val)
            print suffix + ":", md5("|".join(vals)).hexdigest()

    print suffix + ":", time.time() - t