Sort input to krakenuniq to enable retrieval of cached batch runs. #570

CHANGELOG.md

@@ -7,6 +7,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### `Added`
 
+- [#576](https://github.com/nf-core/taxprofiler/pull/576) Sort input to krakenuniq to enable retrieval of cached batch tasks (added by @muniheart)
+
 ### `Fixed`
 
 - [573](https://github.com/nf-core/taxprofiler/pull/573) Improved help messages and documentation to state many of the taxpasta related params require taxonomy files to be input (❤️ to @alexhbnr for reporting, fix by @jfy133)

conf/modules.config

@@ -603,6 +603,7 @@ process {
     }
 
     withName: KRAKENUNIQ_PRELOADEDKRAKENUNIQ {
+        tag = { "${meta.db_name}|${task.index}" }
         ext.args = { "${meta.db_params}" }
         // one run with multiple samples, so fix ID to just db name to ensure clean log name
         ext.prefix = { "${meta.db_name}.krakenuniq" }

subworkflows/local/profiling.nf

@@ -384,7 +384,8 @@ workflow PROFILING {
     }
 
     if ( params.run_krakenuniq ) {
-        ch_input_for_krakenuniq =  ch_input_for_profiling.krakenuniq
+
+        ch_input_for_krakenuniq = ch_input_for_profiling.krakenuniq
             .map {
                 meta, reads, db_meta, db ->
                     def seqtype = (reads[0].name ==~ /.+?\.f\w{0,3}a(\.gz)?$/) ? 'fasta' : 'fastq'
@@ -395,6 +396,9 @@ workflow PROFILING {
             }
             .groupTuple(by: [0,2,3])
             .flatMap { single_meta, reads, db_meta, db ->
+                // Sort reads array by comparing last element, prefix. This will ensure batch membership remains
+                // constant across runs, enabling retrieval of cached tasks.
+                reads.sort { a,b -> a[-1] <=> b[-1] }
                 def batches = reads.collate(params.krakenuniq_batch_size)
                 return batches.collect { batch ->
                     // We split the sample identifier from the reads again after batching.