updates to amplicon and metagenomics validation

Author: AstrobioMike

GL-validate-amplicon

@@ -208,7 +208,9 @@ def check_multiqc_outputs(sample_names):
     """ makes sure all samples' read files are in the multiqc outputs """
 
     # checking raw
-    zip_file = zipfile.ZipFile(fastqc_dir + str(args.output_prefix) + "raw_multiqc_data.zip")
+    raw_multiqc_data_path = fastqc_dir + str(args.output_prefix) + "raw_multiqc_data.zip"
+    check_for_file_and_contents(raw_multiqc_data_path)
+    zip_file = zipfile.ZipFile(raw_multiqc_data_path)
     df = pd.read_csv(zip_file.open("multiqc_general_stats.txt"), sep = "\t", usecols = ["Sample"])
     file_prefixes_in_multiqc = df["Sample"].tolist()
 
@@ -232,7 +234,9 @@ def check_multiqc_outputs(sample_names):
                 report_failure("The raw multiqc output is missing the expected '" + sample + suffix + "' entry.")
 
     # checking filtered
-    zip_file = zipfile.ZipFile(fastqc_dir + str(args.output_prefix) + "filtered_multiqc_data.zip")
+    filt_multiqc_data_path = fastqc_dir + str(args.output_prefix) + "filtered_multiqc_data.zip"
+    check_for_file_and_contents(filt_multiqc_data_path)
+    zip_file = zipfile.ZipFile(filt_multiqc_data_path)
     df = pd.read_csv(zip_file.open("multiqc_general_stats.txt"), sep = "\t", usecols = ["Sample"])
     file_prefixes_in_multiqc = df["Sample"].tolist()
 

GL-validate-metagenomics

@@ -81,7 +81,7 @@ processing_tar_file = "processing_info.tar"
 expected_tar_contents = ["/Snakefile", "/config.yaml", "/envs", "/logs", "/scripts", "/unique-sample-IDs.txt"]
 
 expected_log_file_suffixes = ["-CAT.log", "-assembly.log", "-bam-summarize-and-metabat.log", "-bowtie2-build.log", 
-                              "-kofamscan.log", "-pileup.log", "-prodigal.log", "-humann3-run.log"]
+                              "-bbduk.log", "-kofamscan.log", "-pileup.log", "-prodigal.log", "-humann3-run.log"]
 
 
 validation_log = str(args.GLDS_ID) + "-metagenomics-validation.log"
@@ -541,11 +541,6 @@ def check_processing_tar(samples):
             if target_log not in entries:
                 report_failure("The '" + str(processing_tar_file) + "' does not have the '" + str(target_log) + "' log file as expected.")
 
-        target_log = str(logs_dir) + "bbduk-" + str(sample) + ".log"
-
-        if target_log not in entries:
-            report_failure("The '" + str(processing_tar_file) + "' does not have the '" + str(target_log) + "' log file as expected.")
-
 
 if __name__ == "__main__":
     main()

GL-version

@@ -1,6 +1,6 @@
 #!/usr/bin/env bash
 
-version='1.0.13'
+version='1.0.14'
 
 GREEN='\033[0;32m'
 NC='\033[0m'