dev

musical/__init__.py

@@ -2,7 +2,7 @@
 """
 
 from .utils import beta_divergence
-from .plot import sigplot_bar
+from .plot import sigplot_bar, plot_silhouettes
 from .nmf import NMF
 from .mvnmf import MVNMF, wrappedMVNMF
 from .denovo import DenovoSig
@@ -11,6 +11,7 @@
 
 __all__ = ['beta_divergence',
            'sigplot_bar',
+           'plot_silhouettes',
            'NMF',
            'MVNMF',
            'wrappedMVNMF',

musical/plot.py

@@ -2,8 +2,11 @@
 
 import seaborn as sns
 import numpy as np
+import pandas as pd
 import matplotlib as mpl
 import matplotlib.pyplot as plt
+import matplotlib.gridspec as gridspec
+import matplotlib.ticker as ticker
 from sklearn.preprocessing import normalize
 
 from .utils import snv_types_96_str
@@ -202,3 +205,99 @@ def sigplot_bar(sig, norm=True, figsize=None, title=None,
 
     if outfile is not None:
         plt.savefig(outfile, bbox_inches='tight')
+
+
+def plot_silhouettes(model,  title_tag=None, plotpvalues=True,
+                tick_fontsize=12, label_fontsize=14,
+                outfile=None):
+
+    """Plotting function for silhouette scores.
+
+    model is a MuSiCal object that has been processed with model.fit().
+
+    Parameters
+    ----------
+    model : MuSiCal object after fitting using NMF or mnNMF
+
+    title_tag : string
+        String to add to plot titles, e.g. disease type
+    """
+
+    #Convert dictionaries of all mean silhouette scores and all reconstruction errors to arrays
+    sil_score_mean_array = np.array(list(model.sil_score_mean_all.values()))
+    reconstruction_error_array = np.array(list(model.reconstruction_error_all.values()))
+
+    #Create DF from silhouette scores for heatmap and rename columns
+    sil_score_all_df = pd.DataFrame.from_dict(model.sil_score_all,orient = 'index')
+    sil_score_all_df.columns=[i for i in range(1, sil_score_all_df.shape[1] + 1)]
+
+    #Plot
+
+    # Set up the axes with gridspec
+    fig = plt.figure(figsize=(16, 4))
+    grid = plt.GridSpec(1, 5, hspace=0.5, wspace=4)
+    host = fig.add_subplot(grid[0, 0:3])
+    plt2 = host.twinx()
+
+    if plotpvalues:
+        plt3 = host.twinx()
+
+    heat_map = fig.add_subplot(grid[0, 3:])
+
+    #Generate line plot  
+    host.set_xlabel("n components")
+    host.set_ylabel("Mean silhouette score")
+    plt2.set_ylabel("Reconstruction error")
+    if plotpvalues:
+        plt3.set_ylabel("p-value")
+
+    color1 = '#E94E1B'
+    color2 = '#1D71B8'
+    if plotpvalues:
+        color3 = '#2FAC66'
+
+    p1, = host.plot(model.n_components_all, sil_score_mean_array, color=color1, label="Mean silhouette score", linestyle='--', marker='o',)
+    p2, = plt2.plot(model.n_components_all, reconstruction_error_array,    color=color2, label="Reconstruction error", linestyle=':', marker='D')
+    if plotpvalues:
+        p3, = plt3.plot(model.n_components_all[1:], model.pvalue_all,    color=color3, label="p-value", linestyle='-.', marker='.', alpha=0.5)
+    if plotpvalues:
+        lns = [p1, p2, p3]
+    else:
+        lns = [p1, p2]
+
+    host.legend(handles=lns, loc="lower center", bbox_to_anchor=(0.5, -0.5))
+
+    host.yaxis.label.set_color(p1.get_color())
+    plt2.yaxis.label.set_color(p2.get_color())
+
+    if plotpvalues:
+        plt3.yaxis.label.set_color(p3.get_color())
+
+    #Adjust p-value spine position
+    if plotpvalues:
+        plt3.spines['right'].set_position(('outward', 65))
+
+    #Set ticks interval to 1
+    host.xaxis.set_major_locator(ticker.MultipleLocator(1))
+
+    #Higlight suggested signature
+    host.axvspan(model.n_components-0.25, model.n_components+0.25, color='grey', alpha=0.3)
+
+    #Set title
+    if title_tag is not None:
+        host.set_title('Silhouette scores and reconstruction errors for '+title_tag)
+    else:
+        host.set_title('Silhouette scores and reconstruction errors')
+
+    #Generate heatmap
+    heat_map = sns.heatmap(sil_score_all_df,vmin=0, vmax=1, cmap="YlGnBu")
+    heat_map.set_xlabel("Signatures")
+    heat_map.set_ylabel("n components")
+
+    if title_tag is not None:
+        heat_map.set_title('Silhouette scores for '+title_tag)
+    else:
+        heat_map.set_title('Silhouette scores')
+
+    if outfile is not None:
+        plt.savefig(outfile, bbox_inches='tight')