feat: Add filtering based on N50

- Static initially
beiko-lab · Sep 4, 2023 · 4e8649e · 4e8649e
1 parent 59bfa75
commit 4e8649e
Show file tree

Hide file tree

Showing 2 changed files with 31 additions and 1 deletion.
diff --git a/subworkflows/local/assemblyqc.nf b/subworkflows/local/assemblyqc.nf
@@ -42,7 +42,27 @@ workflow CHECK_ASSEMBLIES {
         ch_software_versions = ch_software_versions.mix(QUAST.out.versions.ifEmpty(null))
         ch_multiqc_files = ch_multiqc_files.mix(QUAST.out.tsv.collect())
 
+        filtered_assemblies = assemblies
+
+        if (params.apply_filtering) {
+            println 'WARNING: Your assemblies are being filtered \n(--apply_filtering is true)'
+
+            QUAST.out.transposed_report
+                .splitCsv(header: true, sep: '\t')
+                .filter { row -> row.N50.toFloat() > 100000}
+                .map { row -> row.Assembly }
+                .set { to_keep }
+
+            filtered_assemblies
+                .combine (to_keep.collect().map { [it] })
+                .filter { meta, path, to_keep -> (path.getSimpleName() in to_keep) }
+                .map { it[0, 1] }
+                .set { filtered_assemblies }
+
+        }
+
     emit:
+        filtered_assemblies
         quast_report = QUAST.out.transposed_report
         assemblyqc_software = ch_software_versions
         multiqc = ch_multiqc_files

diff --git a/workflows/arete.nf b/workflows/arete.nf
@@ -153,6 +153,11 @@ workflow ARETE {
     )
     ch_software_versions = ch_software_versions.mix(CHECK_ASSEMBLIES.out.assemblyqc_software)
 
+    if (params.apply_filtering) {
+        CHECK_ASSEMBLIES.out.filtered_assemblies
+            .set { assemblies }
+    }
+
     if (db_cache) {
         /////////////////// ANNOTATION ///////////////////////////
         ANNOTATE_ASSEMBLIES(
@@ -365,14 +370,19 @@ workflow ANNOTATION {
 
     ANNOTATION_INPUT_CHECK.out.genomes.set { assemblies }
 
-    if (params.run_recombination) {
+    if (params.run_recombination || params.apply_filtering) {
         CHECK_ASSEMBLIES(
             assemblies,
             ch_reference_genome,
             use_reference_genome
         )
         ch_software_versions = ch_software_versions.mix(CHECK_ASSEMBLIES.out.assemblyqc_software)
         ch_multiqc_files = ch_multiqc_files.mix(CHECK_ASSEMBLIES.out.multiqc)
+
+        if (params.apply_filtering) {
+            CHECK_ASSEMBLIES.out.filtered_assemblies
+                .set { assemblies }
+        }
     }