fix --run_ancestry variant verify

PGScatalog · Oct 10, 2024 · 66373b1 · 66373b1
1 parent 676030c
commit 66373b1
Show file tree

Hide file tree

Showing 2 changed files with 26 additions and 8 deletions.
diff --git a/modules/local/ancestry/relabel_scorefiles.nf b/modules/local/ancestry/relabel_scorefiles.nf
@@ -3,7 +3,7 @@ process RELABEL_SCOREFILES {
     label 'process_medium'
     label 'pgscatalog_utils' // controls conda, docker, + singularity options
 
-    tag "$meta.id $meta.effect_type $target_format"
+    tag "reference $meta.effect_type $target_format"
 
     conda "${task.ext.conda}"
 
@@ -16,13 +16,12 @@ process RELABEL_SCOREFILES {
     tuple val(meta), path(target), path(matched)
 
     output:
-    tuple val(relabel_meta), path("${output}"), emit: relabelled
+    tuple val(relabel_meta), path("reference*"), emit: relabelled
     path "versions.yml", emit: versions
 
     script:
     target_format = target.getName().tokenize('.')[1] // test.tar.gz -> tar, test.var -> var
     relabel_meta = meta.plus(['target_format': target_format]) // .plus() returns a new map
-    output_mode = "--split --combined" // always output split and combined data to make life easier
     col_from = "ID_TARGET"
     col_to = "ID_REF"
     output = "${meta.id}.${target_format}*"
@@ -32,11 +31,18 @@ process RELABEL_SCOREFILES {
         --col_to $col_to \
         --target_file $target \
         --target_col ID \
-        --dataset ${meta.id}.${target_format} \
+        --dataset reference \
         --verbose \
-        $output_mode \
+        --combined \
         --outdir \$PWD
 
+    # TODO: improve pgscatalog-relabel so you can set output names precisely
+    # use some unpleasant sed to keep a consistent naming scheme
+    # hgdp_ALL_additive_0.scorefile.gz -> reference_ALL_additive_0.scorefile.gz 
+    output=\$(echo $target | sed 's/^[^_]*_/reference_/')
+
+    mv reference_ALL_relabelled.gz \$output
+
     cat <<-END_VERSIONS > versions.yml
     ${task.process.tokenize(':').last()}:
         pgscatalog.core: \$(echo \$(python -c 'import pgscatalog.core; print(pgscatalog.core.__version__)'))

diff --git a/subworkflows/local/apply_score.nf b/subworkflows/local/apply_score.nf
@@ -125,18 +125,30 @@ workflow APPLY_SCORE {
 
     // pgscatalog-aggregate --verify_variants notes:
     // Checks that variant IDs in the scorefiles match the IDs of scored variants perfectly
-    // Just dump all of the supporting files into the same directory: don't do any fancy joins
+    // Just dump all of the supporting files into the same directory: don't do any fancy channel manipulation
     PLINK2_SCORE.out.vars_scored
         .collect()
         .set { ch_vars_scored }
 
     ch_target_scorefile.flatMap { it.last() }
         .filter(Path)
         .collect()
-        .set{ ch_scorefile_verify }
+        .set{ ch_target_scorefile_flat }
 
-    SCORE_AGGREGATE ( ch_scores, ch_vars_scored, ch_scorefile_verify )
+    // note, for the calculated score:
+    // reference_ALL_additive_0.sscore.zst (ch_scores)
+    // --verify_variants expects the following files in the same directory
+    // reference_ALL_additive_0.sscore.vars (ch_vars_scored)
+    // reference_ALL_additive_0.scorefile.gz (ch_verify_scorefiles)
+
+    ch_apply_ref.flatMap { it.last() }
+        .filter(Path)
+        .mix( ch_target_scorefile_flat )
+        .collect()
+        .set{ ch_verify_scorefiles }
 
+    SCORE_AGGREGATE ( ch_scores, ch_vars_scored, ch_verify_scorefiles )
+
     ch_versions = ch_versions.mix(SCORE_AGGREGATE.out.versions)
 
     emit: