work on consistency check

BCCDC-PHL · Jun 11, 2024 · fda31bb · fda31bb
1 parent 1f40b0d
commit fda31bb
Showing 1 changed file with 58 additions and 17 deletions.
diff --git a/.github/scripts/check_outputs.py b/.github/scripts/check_outputs.py
@@ -93,41 +93,80 @@ def main(args):
             'HA_contigs':             {"upstream": os.path.join(sample_id, "HA_contigs.fa"),
                                        "origin":   os.path.join(sample_id,
                                                                 "analysis_by_stage",
-                                                                "01_assemble_contigs",
+                                                                "03_scaffolding",
                                                                 "HA_contigs.fa")},
             'HA_contigs_alignment':   {"upstream": os.path.join(sample_id, "HA_contigs.afa"),
                                        "origin":   os.path.join(sample_id,
                                                                 "analysis_by_stage",
-                                                                "01_assemble_contigs",
+                                                                "03_scaffolding",
                                                                 "HA_contigs.afa")},
             'NA_contigs':             {"upstream": os.path.join(sample_id, "NA_contigs.fa"),
-                                       "origin":   os.path.join(sample_id, "NA_contigs.fa")},
+                                       "origin":   os.path.join(sample_id,
+                                                                "analysis_by_stage",
+                                                                "03_scaffolding",
+                                                                "NA_contigs.fa")},
             'NA_contigs_alignment':   {"upstream": os.path.join(sample_id, "NA_contigs.afa"),
-                                       "origin":   os.path.join(sample_id, "NA_contigs.afa")},
+                                       "origin":   os.path.join(sample_id,
+                                                                "analysis_by_stage",
+                                                                "03_scaffolding",
+                                                                "NA_contigs.afa")},
             'NP_contigs':             {"upstream": os.path.join(sample_id, "NP_contigs.fa"),
-                                       "origin":   os.path.join(sample_id, "NP_contigs.fa")},
+                                       "origin":   os.path.join(sample_id,
+                                                                "analysis_by_stage",
+                                                                "03_scaffolding",
+                                                                "NP_contigs.fa")},
             'NP_contigs_alignment':   {"upstream": os.path.join(sample_id, "NP_contigs.afa"),
-                                       "origin":   os.path.join(sample_id, "NP_contigs.afa")},
+                                       "origin":   os.path.join(sample_id,
+                                                                "analysis_by_stage",
+                                                                "03_scaffolding",
+                                                                "NP_contigs.afa")},
             'PA_contigs':             {"upstream": os.path.join(sample_id, "PA_contigs.fa"),
-                                       "origin":  os.path.join(sample_id, "PA_contigs.fa")},
+                                       "origin":  os.path.join(sample_id,
+                                                               "analysis_by_stage",
+                                                               "03_scaffolding",
+                                                               "PA_contigs.fa")},
             'PA_contigs_alignment':   {"upstream": os.path.join(sample_id, "PA_contigs.afa"),
-                                       "origin":   os.path.join(sample_id, "PA_contigs.afa")},
+                                       "origin":   os.path.join(sample_id,
+                                                                "analysis_by_stage",
+                                                                "03_scaffolding",
+                                                                "PA_contigs.afa")},
             'PB1_contigs':            {"upstream": os.path.join(sample_id, "PB1_contigs.fa"),
-                                        "origin":   os.path.join(sample_id, "PB1_contigs.fa")},
+                                       "origin":   os.path.join(sample_id,
+                                                                "analysis_by_stage",
+                                                                "03_scaffolding",
+                                                                "PB1_contigs.fa")},
             'PB1_contigs_alignment':  {"upstream": os.path.join(sample_id, "PB1_contigs.afa"),
-                                       "origin":   os.path.join(sample_id, "PB1_contigs.afa")},
+                                       "origin":   os.path.join(sample_id,
+                                                                "analysis_by_stage",
+                                                                "03_scaffolding",
+                                                                "PB1_contigs.afa")},
             'PB2_contigs':            {"upstream": os.path.join(sample_id, "PB2_contigs.fa"),
-                                       "origin":   os.path.join(sample_id, "PB2_contigs.fa")},
+                                       "origin":   os.path.join(sample_id,
+                                                                "analysis_by_stage",
+                                                                "03_scaffolding",
+                                                                "PB2_contigs.fa")},
             'PB2_contigs_alignment':  {"upstream": os.path.join(sample_id, "PB2_contigs.afa"),
-                                       "origin":   os.path.join(sample_id, "PB2_contigs.afa")},
+                                       "origin":   os.path.join(sample_id,
+                                                                "analysis_by_stage",
+                                                                "03_scaffolding",
+                                                                "PB2_contigs.afa")},
             'normalized_reads_r1':    {"upstream": os.path.join(sample_id, "R1.fq"),
-                                       "origin":   os.path.join(sample_id, "R1.fq")},
+                                       "origin":   os.path.join(sample_id,
+                                                                "analysis_by_stage",
+                                                                "00_normalize_depth",
+                                                                f"{sample_id}-normalized_R1.fastq.gz")},
             'normalized_reads_r2':    {"upstream": os.path.join(sample_id, "R2.fq"),
-                                       "origin":   os.path.join(sample_id, "R2.fq")},
+                                       "origin":   os.path.join(sample_id,
+                                                                "analysis_by_stage",
+                                                                "00_normalize_depth",
+                                                                f"{sample_id}-normalized_R2.fastq.gz")},
             'alignment_sam':          {"upstream": os.path.join(sample_id, "alignment.sam"),
                                        "origin":   os.path.join(sample_id, "alignment.sam")},
             'ambig_tsv':              {"upstream": os.path.join(sample_id, "ambig.tsv"),
-                                       "origin":   os.path.join(sample_id, "ambig.tsv")},
+                                       "origin":   os.path.join(sample_id,
+                                                                "analysis_by_stage",
+                                                                "",
+                                                                "ambig.tsv")},
             'contigs_blast':          {"upstream": os.path.join(sample_id, "contigs_blast.tsv"),
                                        "origin":   os.path.join(sample_id, "contigs_blast.tsv")},
             'depth_of_cov_freebayes': {"upstream": os.path.join(sample_id, "depth_of_cov_freebayes.tsv"),
@@ -172,6 +211,8 @@ def main(args):
     all_expected_files_exist = all([check['upstream_exists'] and check['origin_exists'] for check in expected_files_exist_checks])
 
     files_whose_md5sums_are_not_expected_to_match = [
+        'normalized_reads_r1',
+        'normalized_reads_r2',
         'alignment_sam',
         'pileup_vcf',
     ]
@@ -220,10 +261,10 @@ def main(args):
         writer = csv.DictWriter(f, fieldnames=output_fields, extrasaction='ignore')
         writer.writeheader()
         for test in tests:
+            test["test_result"] = "FAIL"
             if test["test_passed"]:
                 test["test_result"] = "PASS"
-            else:
-                test["test_result"] = "FAIL"
+
             writer.writerow(test)
 
     for test in tests: