Clean HelixFold3 outputs

CHANGELOG.md

@@ -98,6 +98,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - [[#313](https://github.com/nf-core/proteinfold/issues/313)] - Harmonize colabfold metrics extraction with other modes.
 - [[#455](https://github.com/nf-core/proteinfold/issues/455)] - Fix colabfold monomer inheriting id from fasta header.
 - [[#457](https://github.com/nf-core/proteinfold/issues/457)] - Fix colabfold multimer always downloading model weights.
+- [[PR #461](https://github.com/nf-core/proteinfold/pulls/461)] - Update publishdir patterns for HelixFold3 module
 - [[PR #464](https://github.com/nf-core/proteinfold/pulls/454)] - Update publishdir patterns for Boltz module
 
 ### Parameters

conf/modules_helixfold3.config

@@ -20,9 +20,7 @@ process {
     }
 
     withName: 'RUN_HELIXFOLD3' {
-
         accelerator = params.use_gpu ?  1 : 0
-
         ext.args = [
             params.helixfold3_max_template_date ? "--max_template_date=${params.helixfold3_max_template_date}" : "--max_template_date=2038-01-19",
             "--model_name allatom_demo",
@@ -32,18 +30,52 @@ process {
             params.helixfold3_precision  ? "--precision ${params.helixfold3_precision}"      : "--precision 'bf16'",
             params.helixfold3_infer_times ? "--infer_times ${params.helixfold3_infer_times}" : "--infer_times 4"
         ].join(' ').trim()
-
         publishDir = [
-            path: { "${params.outdir}/helixfold3/${meta.id}" },
-            mode: 'copy',
-            saveAs: { filename ->
-                if(filename.endsWith('_pae.tsv')){
-                    "paes/$filename"
-                } else if(filename.equals('versions.yml')){
-                    null
-                } else { filename }
-            },
-            pattern: '*.*'
+            [
+                path: { "${params.outdir}/helixfold3/${meta.id}" },
+                mode: 'copy',
+                pattern: '*_plddt.tsv'
+            ],
+            [
+                path: { "${params.outdir}/helixfold3/${meta.id}" },
+                mode: 'copy',
+                pattern: '*_ptm.tsv'
+            ],
+            [
+                path: { "${params.outdir}/helixfold3/${meta.id}" },
+                mode: 'copy',
+                pattern: '*_iptm.tsv'
+            ],
+            [
+                path: { "${params.outdir}/helixfold3/${meta.id}" },
+                mode: 'copy',
+                saveAs: { filename -> "paes/$filename" },
+                pattern: '*_[1-5]_pae.tsv'
+            ],
+            [
+                path: { "${params.outdir}/helixfold3/${meta.id}" },
+                mode: 'copy',
+                pattern: '*_helixfold3_msa.tsv'
+            ],
+            [
+                enabled: params.save_intermediates,
+                path: { "${params.outdir}/helixfold3/${meta.id}/raw" },
+                mode: 'copy',
+                pattern: 'raw/**',
+                saveAs: { filename -> filename.toString().replaceFirst(/^raw\//, '') }
+            ],
+            [
+                path: { "${params.outdir}/helixfold3/top_ranked_structures" },
+                mode: 'copy',
+                saveAs: { "${meta.id}.pdb" },
+                pattern: '*_helixfold3.pdb'
+            ],
+            [
+                path: { "${params.outdir}/helixfold3/top_ranked_structures" },
+                mode: 'copy',
+                saveAs: { "${meta.id}.cif" },
+                pattern: '*_helixfold3.cif'
+            ]
         ]
     }
 

modules/local/run_helixfold3/main.nf

@@ -26,9 +26,10 @@ process RUN_HELIXFOLD3 {
     path ('maxit_src')
 
     output:
+    path ("raw/**")                                         , emit: raw
     tuple val(meta), path ("${meta.id}_helixfold3.pdb")     , emit: top_ranked_pdb
     tuple val(meta), path ("${meta.id}_helixfold3.cif")     , emit: main_cif
-    tuple val(meta), path ("${meta.id}-ranked*.pdb")        , emit: pdb
+    tuple val(meta), path ("raw/${meta.id}-ranked*.pdb")    , emit: pdb
     tuple val(meta), path ("${meta.id}_plddt.tsv")          , emit: multiqc
     tuple val(meta), path ("${meta.id}_helixfold3_msa.tsv") , emit: msa
     // If ${meta.id}-rank*/all_results.json" doesn't have PAE vales in the key, this will be empty
@@ -84,12 +85,13 @@ process RUN_HELIXFOLD3 {
         --pkls "${fasta.baseName}/final_features.pkl" \\
         --jsons ${fasta.baseName}/${fasta.baseName}-rank*/all_results.json
 
-    [ ! -d ${meta.id} ] && mkdir ${meta.id}
+    mkdir -p raw
     for i in 1 2 3 4 5; do
-        cp "${fasta.baseName}/${fasta.baseName}-rank\$i/predicted_structure.pdb" "${meta.id}-ranked_\$i.pdb"
+        cp "${fasta.baseName}/${fasta.baseName}-rank\$i/predicted_structure.pdb" "raw/ranked_\$i.pdb"
     done
 
     mv "${meta.id}_msa.tsv" "${meta.id}_helixfold3_msa.tsv"
+    mv "${fasta.baseName}" raw/
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
@@ -110,12 +112,12 @@ process RUN_HELIXFOLD3 {
     touch "${meta.id}_3_pae.tsv"
     touch "${meta.id}_4_pae.tsv"
     touch "${meta.id}_5_pae.tsv"
-    mkdir "${meta.id}"
-    touch "${meta.id}-ranked_1.pdb"
-    touch "${meta.id}-ranked_2.pdb"
-    touch "${meta.id}-ranked_3.pdb"
-    touch "${meta.id}-ranked_4.pdb"
-    touch "${meta.id}-ranked_5.pdb"
+    mkdir -p raw
+    touch "raw/${meta.id}-ranked_1.pdb"
+    touch "raw/${meta.id}-ranked_2.pdb"
+    touch "raw/${meta.id}-ranked_3.pdb"
+    touch "raw/${meta.id}-ranked_4.pdb"
+    touch "raw/${meta.id}-ranked_5.pdb"
 
 
     cat <<-END_VERSIONS > versions.yml