plots.py 文件源码

def completeness_plot(consensus_data, ordered_genomes, biotypes, completeness_plot_tgt, gene_biotype_map,
                      transcript_biotype_map):
    def adjust_plot(g, gene_count, tx_count):
        for ax, c in zip(*[g.axes[0], [gene_count, tx_count]]):
            _ = ax.set_ylim(0, c)
            ax.spines['top'].set_edgecolor('#e74c3c')
            ax.spines['top'].set_linewidth(2)
            ax.spines['top'].set_visible(True)
            ax.spines['top'].set_linestyle('dashed')

    df = json_grouped_biotype_nested_counter_to_df(consensus_data, 'Completeness')
    with completeness_plot_tgt.open('w') as outf, PdfPages(outf) as pdf:
        tot_df = df.groupby(by=['genome', 'category']).aggregate(np.sum).reset_index()
        tot_df = sort_long_df(tot_df, ordered_genomes)
        title = 'Number of comparative genes/transcripts present'
        g = generic_barplot(pdf=pdf, data=tot_df, x='genome', y='count', col='category', xlabel='',
                            sharey=False, ylabel='Number of genes/transcripts', title=title,
                            col_order=['Gene', 'Transcript'], close=False, palette=choose_palette(ordered_genomes))
        adjust_plot(g, len(gene_biotype_map), len(transcript_biotype_map))
        multipage_close(pdf, tight_layout=False)
        for biotype in biotypes:
            biotype_df = biotype_filter(df, biotype)
            if biotype_df is not None:
                biotype_df = sort_long_df(biotype_df, ordered_genomes)
                gene_biotype_count = len({i for i, b in gene_biotype_map.iteritems() if b == biotype})
                tx_biotype_count = len({i for i, b in transcript_biotype_map.iteritems() if b == biotype})
                title = 'Number of comparative genes/transcripts present for biotype {}'.format(biotype)
                g = generic_barplot(pdf=pdf, data=biotype_df, x='genome', y='count', col='category', xlabel='',
                                    sharey=False, ylabel='Number of genes/transcripts',
                                    title=title, col_order=['Gene', 'Transcript'], close=False,
                                    palette=choose_palette(ordered_genomes))
                adjust_plot(g, gene_biotype_count, tx_biotype_count)
                multipage_close(pdf, tight_layout=False)