nanoQC.py 文件源码

def plot_nucleotide_diversity(ax, fqlists, invert=False):
    '''
    Create a FastQC-like "?Per base sequence content" plot
    Plot fraction of nucleotides per position
    zip will stop when shortest read is exhausted
    '''
    if invert:
        fqlists = [list(reversed(read)) for read in fqlists]
    numreads = len(fqlists)
    sns.set_style("darkgrid")
    l_A, = ax.plot(
        np.array([pos.count('A') / numreads for pos in zip(*fqlists)]), 'green', label='A')
    l_T, = ax.plot(
        np.array([pos.count('T') / numreads for pos in zip(*fqlists)]), 'red', label='T')
    l_G, = ax.plot(
        np.array([pos.count('G') / numreads for pos in zip(*fqlists)]), 'black', label='G')
    l_C, = ax.plot(
        np.array([pos.count('C') / numreads for pos in zip(*fqlists)]), 'blue', label='C')
    if invert:
        ax.set_xticklabels(-1 * ax.get_xticks().astype(int))
    return [l_A, l_T, l_G, l_C]