genbank.py 文件源码

def getUniqueRedundSets(fileName,speciesName):
    '''Run through genbank file fileName, get set of unique genes (with no
redundancies), and set with redundancies.'''
    f = open(fileName, 'rU')
    geneL=[]
    for record in SeqIO.parse(f, "genbank"):
        # iterate through the genes on the chromosome
        for feature in record.features:
            # choose only the features that are protein coding genes
            if feature.type == "CDS" and 'protein_id' in feature.qualifiers:
                geneName = speciesName + '-' + feature.qualifiers['protein_id'][0]
                geneL.append(geneName)

    f.close()

    # now figure out which ones are unique
    uniqueS = set()
    redundS = set()
    for gene in geneL:
        if geneL.count(gene)>1:
            redundS.add(gene)
        else:
            uniqueS.add(gene)
    return uniqueS,redundS