filter_EVM_RNASeq.py

import subprocess

run_command = 'find . -maxdepth 2 -name evm.out'
process = subprocess.Popen(run_command, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=True)
for c in iter(lambda: process.stdout.readline(), ''):
    out = open(c.rstrip().replace('.out','_nonRNARemoved.out'), 'w')
    this_gene = []
    for line in open(c.rstrip()):
        if line.startswith('!!'):
            out.write(line)
            continue
        if line.startswith('#'):
            if this_gene:
                # check whether this_gene has RNASeq in it
                keep = False
                for l in this_gene:
                    if 'Cufflinks' in l:
                        keep = True
                        break

                if not keep: continue

                for l in this_gene:
                    out.write(l)

            this_gene = [line]
        else:
            this_gene.append(line)
    # the last gene still has to be printed
    if this_gene:
        keep = False
        for l in this_gene:
            if 'Cufflinks' in l:
                keep = True
                break

        if not keep: continue

        for l in this_gene:
            out.write(l)