Updated reports.

scripts/ReportSizes.py

@@ -20,11 +20,11 @@ def make_report(size_gistic, size_rubic_gains, size_rubic_losses, census_genes,
         for tool in 'GISTIC', 'RUBIC':
             size_file = (size_rubic_gains[i], size_rubic_losses[i]) if tool == 'RUBIC' else size_gistic[i]
             parsed_results = parse_regions(size_file, known_genes, census_genes, tool)
-            if tool not in all_results.keys():
-                all_results[tool] = {}
-                all_results[tool][size] = [parsed_results]
-            else:
-                all_results[tool][size] = all_results[tool][size] + [parsed_results]
+            #if tool not in all_results.keys():
+            #    all_results[tool] = {}
+            #    all_results[tool][size] = [parsed_results]
+            #else:
+            #    all_results[tool][size] = all_results[tool][size] + [parsed_results]
             stats_results = get_stats(parsed_results, size)
             for stat_list in stats_results[0], stats_results[1]:
                 converted_stats = [tool] + stat_list[0:2]
@@ -33,7 +33,7 @@ def make_report(size_gistic, size_rubic_gains, size_rubic_losses, census_genes,
                 for stat in stat_list[5:]:
                     converted_stats.append(float(stat.split(" (")[0]))
                 list_stats.append(converted_stats)
-    overlap_genes(all_results, report_file)
+    #overlap_genes(all_results, report_file)
     make_plots(list_stats, reps, sizes, plot_dir)
 
 def overlap_genes(all_results, report_file):
@@ -60,7 +60,7 @@ def plot_size_differences(df_stats, value_y_axis, list_sizes, nr_reps, plot_dir)
     """Plot the differences between analyses using different sample sizes."""
     sns.set_style("whitegrid")
     g = sns.factorplot(x="Sample size", y=value_y_axis, col="Type", hue="Tool", data=df_stats, kind="box",
-                       size=5, aspect=1, palette=sns.cubehelix_palette(8, start=.5, rot=-.75, dark=.2))
+                       size=5, aspect=1, palette=["#5975A4","#5F9E6E"])
     g.set_axis_labels("Sample size", value_y_axis).set_titles("{col_name}").despine(bottom=True)
     #add_significance(g, df_stats, list_sizes, value_y_axis)
     png_file = os.path.join(plot_dir, value_y_axis.replace(" ", "_") + ".png")

scripts/ReportTools.py

@@ -24,11 +24,11 @@ def make_report(gistic_results, rubic_gain_results, rubic_loss_results,
         parsed.append(parsed_results)
         stats.append(get_stats(parsed_results, tool))
     stats.append(parse_bed_overlap(bed_overlap, known_genes, census_genes))
-    table_all_regions(parsed, file_regions)     #make table with all regions
-    make_tool_report(file_tools, stats)     #make tool report
+    table_all_regions(parsed, file_regions)                                 #make table with all regions
+    make_tool_report(file_tools, stats)                                     #make tool report
     all_genes = make_gene_files(parsed, file_genes_GISTIC, file_genes_RUBIC, file_genes_both, known_genes, file_venn)    #make gene files and venn overlap plot
-    plot_histogram_sizes(parsed, file_swarmplot)    #plot histogram with the sizes of the regions
-    make_bed_known_regions(known_genes, biomart_file, bed_known_genes)     #Make bed file with the locations of known genes.
+    plot_histogram_sizes(parsed, file_swarmplot)                            #plot histogram with the sizes of the regions
+    make_bed_known_regions(known_genes, biomart_file, bed_known_genes)      #Make bed file with the locations of known genes.
 
 def make_tool_report(file_tools, stats_tools):
     """Make a report of the results from GISTIC and RUBIC."""
@@ -86,6 +86,8 @@ def make_gene_file_overlap(gene_lists, out_both):
 def venn3_overlap(gene_lists, known_genes, out_venn):
     known = parse_gene_file(known_genes, np.inf)
     plt.subplots(figsize=(7, 7))
+    overlap_tools = set(gene_lists[0]).intersection(set(gene_lists[1]))
+    print(overlap_tools.intersection(set(known)))
     venn_sets = [set(gene_lists[0]), set(gene_lists[1]), set(known)]
     c = venn3(venn_sets, ('GISTIC2', 'RUBIC', '                 Known genes'))
     c.get_patch_by_id('100').set_color('#5975A4'), c.get_patch_by_id('010').set_color('#5F9E6E')