updated dragonn workshop tutorial

kundajelab · Aug 24, 2016 · cf033e3 · cf033e3
1 parent e5f5ff8
commit cf033e3
Show file tree

Hide file tree

Showing 2 changed files with 1,210 additions and 239 deletions.
diff --git a/dragonn/tutorial_utils.py b/dragonn/tutorial_utils.py
@@ -185,21 +185,16 @@ def plot_SequenceDNN_layer_outputs(dnn, simulation_data):
         ax2.axvspan(conv_output_start, conv_output_stop, color='grey', alpha=0.5)
 
 
-def interpret_SequenceDNN_distributed(dnn, simulation_data, plot_layer_outputs=False):
+def interpret_SequenceDNN_filters(dnn, simulation_data):
     print("Plotting simulation motifs...")
     plot_motifs(simulation_data)
     plt.show()
     print("Visualizing convolutional sequence filters in SequenceDNN...")
     plot_sequence_filters(dnn)
     plt.show()
-    if plot_layer_outputs:
-        print("%s %s" % ("Plotting outputs of convolutional and max pooling layer",
-                         "for a positive simulation example..."))
-        plot_SequenceDNN_layer_outputs(dnn, simulation_data)
-        plt.show()
 
 
-def interpret_SequenceDNN_integrative(dnn, simulation_data):
+def interpret_data_with_SequenceDNN(dnn, simulation_data):
     # get a positive and a negative example from the simulation data
     pos_indx = np.where(simulation_data.y_valid==1)[0][0]
     pos_X = simulation_data.X_valid[pos_indx:(pos_indx+1)]
@@ -277,19 +272,21 @@ def interpret_SequenceDNN_integrative(dnn, simulation_data):
                 if score_type=='Motif Scores':
                     scores_to_plot = scores[0, _i, :]
                 else:
-                    scores_to_plot = scores.squeeze(axis=2)
+                    scores_to_plot = scores[0, 0, 0, :]
                 if motif_label not in motif_labels_cache:
                     motif_labels_cache.append(motif_label)
                     add_legend = True
                 motif_color = motif_colors[motif_labels_cache.index(motif_label)]
                 ax.plot(scores_to_plot, label=motif_label, c=motif_color)
-                if add_legend:
-                    leg = ax.legend(loc=[0,0.85], frameon=False, fontsize=font_size,
-                                    ncol=3, handlelength=-0.5)
-                    for legobj in leg.legendHandles:
-                        legobj.set_color('w')
-                    for _i, text in enumerate(leg.get_texts()):
-                        text.set_color(motif_color)
+            if add_legend:
+                leg = ax.legend(loc=[0,0.85], frameon=False, fontsize=font_size,
+                                ncol=3, handlelength=-0.5)
+                for legobj in leg.legendHandles:
+                    legobj.set_color('w')
+                for _j, text in enumerate(leg.get_texts()):
+                    text_color = motif_colors[
+                        motif_labels_cache.index(motif_label_dict[score_type][_j])]
+                    text.set_color(text_color)
             for motif_site in motif_sites[key]:
                 ax.axvspan(motif_site - highlight_width, motif_site + highlight_width,
                            color='grey', alpha=0.1)