Bloguero_Connor

from Bio import SeqIO
from Bio.Blast import NCBIXML

def blastcomp(fastafile,blastfile):
    handle = open(fastafile)
    fastanames=set()
    #Reads the fasta names
    for record in SeqIO.parse(handle, "fasta") :
        fastanames.add(record.name)
    handle.close()
    blastnames=set()
    #Reads the blast names
    b_records=NCBIXML.parse(open(blastfile))
    for b_record in b_records:
        blastnames.add(b_record.query)
    return fastanames.difference(blastnames)


blastfile="/home/sbassi/bioinfo/INTA/filtracMT.xml"
fastafile='INTA/allfiltrados.txt'
print blastcomp(fastafile,blastfile)