models.py 文件源码

def generate_corpusfile(corpus_fname, n, out):
    '''
    Args:
        corpus_fname: corpus file name
        n: the number of chunks to split. In other words, "n" for "n-gram"
        out: output corpus file path
    Description:
        Protvec uses word2vec inside, and it requires to load corpus file
        to generate corpus.
    '''
    f = open(out, "w")
    for r in SeqIO.parse(corpus_fname, "fasta"):
        ngram_patterns = split_ngrams(r.seq, n)
        for ngram_pattern in ngram_patterns:
            f.write(" ".join(ngram_pattern) + "\n")
        sys.stdout.write(".")

    f.close()