filter.py 文件源码

def filtered_table(table,
        v_gene_coverage,  # at least
        j_gene_coverage,  # at least
        v_gene_evalue,  # at most
    ):
    """
    Discard the following rows in the table:
    - no J assigned
    - stop codon found
    - V gene coverage less than v_gene_coverage
    - J gene coverage less than j_gene_coverage
    - V gene E-value greater than v_gene_evalue

    Return the filtered table.
    """
    stats = FilteringStatistics()
    stats.n = len(table)
    # Both V and J must be assigned
    # (Note V_gene and J_gene columns use empty strings instead of NA)
    filtered = table[(table['V_gene'] != '') & (table['J_gene'] != '')][:]
    stats.vjassigned = len(filtered)
    filtered['V_gene'] = pd.Categorical(filtered['V_gene'])

    # Filter out sequences that have a stop codon
    filtered = filtered[filtered.stop == 'no']
    stats.stop = len(filtered)

    # Filter out sequences with a too low V gene hit E-value
    filtered = filtered[filtered.V_evalue <= v_gene_evalue]
    stats.v_evalue = len(filtered)

    # Filter out sequences with too low V gene coverage
    filtered = filtered[filtered.V_covered >= v_gene_coverage]
    stats.v_coverage = len(filtered)

    # Filter out sequences with too low J gene coverage
    filtered = filtered[filtered.J_covered >= j_gene_coverage]
    stats.j_coverage = len(filtered)

    return filtered, stats