clusterplot.py 文件源码

def plot_clustermap(sequences, title, plotpath, size=300, dpi=200):
    """
    Plot a clustermap of the given sequences

    size -- Downsample to this many sequences
    title -- plot title

    Return the number of clusters.
    """
    logger.info('Clustering %d sequences (downsampled to at most %d)', len(sequences), size)
    sequences = downsampled(sequences, size)
    df, linkage, clusters = cluster_sequences(sequences)

    palette = sns.color_palette([(0.15, 0.15, 0.15)])
    palette += sns.color_palette('Spectral', n_colors=max(clusters), desat=0.9)
    row_colors = [ palette[cluster_id] for cluster_id in clusters ]
    cm = sns.clustermap(df,
            row_linkage=linkage,
            col_linkage=linkage,
            row_colors=row_colors,
            linewidths=None,
            linecolor='none',
            figsize=(210/25.4, 210/25.4),
            cmap='Blues',
            xticklabels=False,
            yticklabels=False
    )
    if title is not None:
        cm.fig.suptitle(title)
    cm.savefig(plotpath, dpi=dpi)

    # free the memory used by the plot
    import matplotlib.pyplot as plt
    plt.close('all')

    return len(set(clusters))