BioCompass.py 文件源码

def get_hits(filename, criteria='cum_BLAST_score'):
    """

        Reproduces original Tiago's code: table_1_extender.py

        In the future allow different criteria. Right now it takes
          from the very first block, which has the highest Cumulative
          BLAST.
    """
    with open(filename) as f:
        df = antiSMASH_to_dataFrame(f.read())

    df.dropna(subset=['query_gene'], inplace=True)
    df.sort_values(by=criteria, ascending=False, na_position='last',
            inplace=True)
    return df.groupby('query_gene', as_index=False).first()