add step 5 to logs

methods/gdc_api_calls.py

@@ -272,6 +272,7 @@ def return_joint_single_cnv_frequency(cnv, cnv_change, cnv_change_5_category):
         if not total_number_of_cases_with_cnv_data:
             continue
 
+        print('\nStep 5: Query GDC and process results\n')
         print('total number of cases with CNV data {}'.format(
                 total_number_of_cases_with_cnv_data))
 
@@ -378,10 +379,6 @@ def get_freq_cnv_loss_or_gain(gene_entities, cancer_entities, query, cnv_and_ssm
                 print("exception: {}".format(str(e)))
                 continue
 
-            # total_number_of_cases_with_cnv_data = get_available_cnv_data_for_project(ce)
-            # skip if cannot obtain total # of cnv cases from API
-            # if not total_number_of_cases_with_cnv_data:
-            #    continue
 
             if not ce in cnv:
                 cnv[ce] = {}
@@ -393,11 +390,8 @@ def get_freq_cnv_loss_or_gain(gene_entities, cancer_entities, query, cnv_and_ssm
                 if item["case"]["case_id"]:
                     case_id_list.append(item["case"]["case_id"])
             number_of_cases_with_cnv_change = len(case_id_list)
-            # freq = number_of_cases_with_cnv_change / total_number_of_cases_with_cnv_data
             cnv[ce][ge]["case_id_list"] = case_id_list
-            # cnv[ce][ge]["frequency"] = round(freq * 100, 2)
 
-    # print('debug: cnv {}'.format(cnv))
     if cnv_and_ssm_flag:
         return cnv
     else:
@@ -458,6 +452,7 @@ def get_msi_frequency(cancer_entities):
             msi_pos = msi_results.count('MSI')
             msi_total = len(msi_results)
             freq = msi_pos / msi_total
+            print('\nStep 5: Query GDC and process results\n')
             print('obtained {} BAM files with MSI tag, out of a total of {} BAM files with MSI information'.format(
                 msi_pos, msi_total
             ))
@@ -588,6 +583,8 @@ def run_cnv_ssm_api(decompose_result, cancer_entities, query):
             ssm_result = get_cases_with_ssms_in_a_gene(
                 project=ce, gene_name=decompose_result["mut_gene"]
             )
+            total_case_count = get_total_variation_data_for_project(project=ce)
+            print('\nStep 5: Query GDC and process results\n')
             # calcuate overlap of cases and return freq
             print('getting shared cases with CNV and SSMs...')
             cases_with_ssm_and_cnvs = [
@@ -595,7 +592,6 @@ def run_cnv_ssm_api(decompose_result, cancer_entities, query):
                 set(ssm_result["case_id_list"]),
             ]
             shared_cases = list(reduce(lambda x, y: x & y, cases_with_ssm_and_cnvs))
-            total_case_count = get_total_variation_data_for_project(project=ce)
             print('number of shared_cases {}'.format(len(shared_cases)))
             print('total case count {}'.format(total_case_count))
             freq = round((len(shared_cases) / total_case_count) * 100, 2)
@@ -634,6 +630,7 @@ def get_top_cases_counts_by_gene(gene_entities, cancer_entities):
                 "cases_without_mutations"
             ] = cases_without_mutations
             top_cases_counts_by_gene[ce]["total_case_count"] = total_case_count
+            print('\nStep 5: Query GDC and process results\n')
             print('obtained {} cases with mutations and a total case count of {}'.format(
                 cases_with_mutations, total_case_count
             ))