screen_dynamics.py 文件源码

def gRNA_rank(s1, s2, prefix="", text=False, n_labels=30):
    # Rank of gRNA change
    fig, axis = plt.subplots(3, 2, sharex=True, sharey=True, figsize=(8, 8))
    axis = axis.flatten()

    for i, screen in enumerate(s2.columns[::-1]):
        x = s1.join(s2)  # .fillna(0)
        x = x.iloc[np.random.permutation(len(x))]

        x = x.ix[x.index[~x.index.str.contains("Wnt")]]
        if prefix.startswith("mid_screen-"):
            b = x["gDNA_Jurkat"]
        else:
            b = x["plasmid_pool_TCR"]

        x = x.fillna(0)
        b = b.fillna(0)

        fc = np.log2(1 + x[screen]) - np.log2(1 + b)

        fc.name = screen
        if i == 0:
            xx = pd.DataFrame(fc)
        else:
            xx = xx.join(fc, how="outer")

        colors = pd.DataFrame()
        colors[sns.color_palette("colorblind")[0]] = x.index.str.contains("Wnt")
        colors[sns.color_palette("colorblind")[1]] = x.index.str.contains("CTRL")
        colors[sns.color_palette("colorblind")[2]] = x.index.str.contains("Tcr")
        colors[sns.color_palette("colorblind")[3]] = x.index.str.contains("Ess")
        colors = colors.apply(lambda x: x[x].index.tolist()[0], axis=1).tolist()

        axis[i].scatter(fc.rank(ascending=False, method="first"), fc, color=colors, alpha=0.5)
        if text:
            for j in x[x.index.str.contains("ETS1|GATA3|RUNX1")].index:
                axis[i].text(
                    fc.rank(ascending=False, method="first").ix[j],
                    fc.ix[j],
                    j)
        axis[i].axhline(y=0, color='black', linestyle='--', lw=0.5)

        axis[i].set_title(screen)

    for i in range(0, len(axis), 2):
        axis[i].set_ylabel("gRNA fold-change")
    for ax in axis[-2:]:
        ax.set_xlabel("gRNA rank")
    sns.despine(fig)
    fig.savefig(os.path.join(results_dir, "gRNA_counts.norm.{}.rank.{}svg".format(prefix, "text." if text else "")), bbox_inches="tight")
    fig.savefig(os.path.join(results_dir, "gRNA_counts.norm.{}.rank.{}pdf".format(prefix, "text." if text else "")), bbox_inches="tight")

    # Save ranked list
    xx.to_csv(os.path.join(results_dir, "gRNA_counts.norm.{}.rank.csv".format(prefix)), index=True)

    # Save ranked list of gene-level measurements, reduced by mean and min
    m = pd.merge(xx.reset_index(), guide_annotation[["oligo_name", "gene"]], left_on="gRNA_name", right_on="oligo_name").drop("oligo_name", axis=1).set_index(["gene", "gRNA_name"])
    m.groupby(level=[0]).mean().to_csv(os.path.join(results_dir, "gRNA_counts.norm.{}.gene_mean.rank.csv".format(prefix)), index=True)
    m.groupby(level=[0]).min().to_csv(os.path.join(results_dir, "gRNA_counts.norm.{}.gene_min.rank.csv".format(prefix)), index=True)