Delete unnecessary BAMs (#147)

* Add file deletion function

* Use deletion function in recalibrate-BAM

* Fix deletion functions

* Update CHANGELOG

* Move changelog item to specific version

* Add comments for functions

* Version bump in nextflow.config

CHANGELOG.md

@@ -15,6 +15,8 @@ This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.htm
 ### Changed
 + Default to alt-aware reference for align-DNA
 + Re-order FASTQ CSV to match order in align-DNA
+### Added
++ Deletion step for normal BAMs when running multi-sample patients in paired mode
 
 ---
 

module/common.nf

@@ -24,3 +24,44 @@ String identify_file(filepath) {
     assert file_found.exists();
     return file_found.toRealPath().toString()
 }
+/**
+*   Function to delete a file once it has been copied over to a final destination
+*/
+void delete_file(String filepath, output_filepattern) {
+    File expected_file = new File(filepath);
+    Integer expected_bytes = expected_file.length(); // The size of the file in bytes
+
+    String output_filepath = '';
+
+    // Wait and find the output file
+    Boolean keep_looking = true;
+    while (keep_looking) {
+        try {
+            // Try to access the final output file
+            output_filepath = identify_file(output_filepattern);
+            System.out.println("Found: ${output_filepath}");
+            keep_looking = false;
+        } catch (AssertionError e) {
+            if (e.toString().replace(' ', '').contains('[]0')) {
+                // Output file not found (empty list of identified files) so wait and retry
+                sleep(5000);
+            } else {
+                // Error with output file existence, skip deletion
+                keep_looking = false;
+                System.out.println("Failed to find final output file: ${output_filepattern}, not deleting ${filepath}.");
+                return;
+            }
+        }
+    }
+
+    File output_file = new File(output_filepath);
+    Integer output_bytes = output_file.length();
+
+    // Wait until final output file matches size in bytes with original file then delete original
+    while (output_bytes != expected_bytes) {
+        sleep(5000);
+        output_bytes = output_file.length();
+    }
+
+    expected_file.delete()
+}

module/recalibrate_BAM/identify_outputs.nf

@@ -1,4 +1,4 @@
-include { identify_file } from '../common'
+include { identify_file; delete_file } from '../common'
 
 workflow identify_recalibrate_bam_outputs {
     take:
@@ -7,9 +7,17 @@ workflow identify_recalibrate_bam_outputs {
     main:
     och_recalibrate_bam.map{ recalibrate_bam_out ->
         recalibrate_bam_out[0].normal.each { normal_sample ->
+            String bam_file = identify_file("${recalibrate_bam_out[1]}/*GATK-*${normal_sample}*.bam");
             if (!params.sample_data[normal_sample]['recalibrate-BAM']['BAM']) {
-                params.sample_data[normal_sample]['recalibrate-BAM']['BAM'] = identify_file("${recalibrate_bam_out[1]}/*GATK-*${normal_sample}*.bam");
+                params.sample_data[normal_sample]['recalibrate-BAM']['BAM'] = bam_file;
                 params.sample_data[normal_sample]['recalibrate-BAM']['contamination_table'] = identify_file("${recalibrate_bam_out[2]}/GATK-*${normal_sample}_alone.table");
+            } else {
+                // Normal file already found so delete any other normals - only triggered when running multiple samples in paired mode
+
+                // Replace the work_dir prefix in the output path with the output_dir prefix for final output
+                String separator = "/recalibrate-BAM-";
+                String dir_stripped = recalibrate_bam_out[1].toString().replaceFirst("${params.work_dir}.*${separator}", "");
+                delete_file(bam_file, "${params.output_dir}/output${separator}${dir_stripped}/*GATK-*${normal_sample}*.bam");
             };
         };
         recalibrate_bam_out[0].tumor.each { tumor_sample ->

nextflow.config

@@ -3,5 +3,5 @@ manifest {
     name = 'metapipeline-DNA'
     author = ['Yash Patel', 'Chenghao Zhu', 'Helena Winata', 'Alfredo Enrique Gonzalez', 'Nicholas Wang', 'Mohammed Faizal Eeman Mootor']
     description = 'Nextflow pipeline to convert BAM to FASTQ, align, call gSNP, call sSNV, call gSV, call sSV and call mtSNV'
-    version = '5.1.0'
+    version = '5.2.0'
 }