merge_suppa.py~

import os, sys

ioeFile = sys.argv[1]

tracker = dict()
out1 = open("merged.%s" % ioeFile,'w') 
with open(ioeFile) as lines:
    print(next(lines).rstrip(),file=out1)
    for l in lines:
        if "seqname" in l: continue
        #chrYENSGR0000002586 ENSGR0000002586;SE:chrY:2594414-2606241:2606297-2608820:+ 999bdcca-a305-418f-93ef-d8818544bffc;0_ENSGR0000002586.13-2,4612f8da-fc52-4967-8e2d-3f75b6095bd0;0_ENSGR0000002586.13fdf42774-d4dd-473a-910d-765ded224db0;16_ENSGR0000002586.13,999bdc
        atype = l.split()[2].split(":")[0].split(";")[-1]
        
        if len(atype) !=2:
            print(l)
        entry = l.rstrip()
        event = entry.split()
        eventData = event[2].replace("-",":").split(":")
        se = (event[0], event[1], eventData[3], eventData[4],atype)
        if se not in tracker:
            tracker[se] = {"alt":set(),"total":set()}
        tracker[se]["alt"].update(list(event[-2].split(",")))
        tracker[se]["total"].update(list(event[-1].split(",")))




for i,k in tracker.items():
    chrom,gene,stuff1,stuff2,stuff3 = i
    print(chrom,gene,"%s:%s-%s;%s" % (chrom,stuff1,stuff2,stuff3),",".join([ i for i in k['alt']]),",".join([i for i in k['total']]),sep="\t", file=out1)

out1.close()