add --verify_variants support

PGScatalog · Oct 9, 2024 · d4fd211 · d4fd211
1 parent d89eb7d
commit d4fd211
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 23 deletions.
diff --git a/modules/local/plink2_score.nf b/modules/local/plink2_score.nf
@@ -64,16 +64,6 @@ process PLINK2_SCORE {
             $input ${geno.baseName} \
             --out ${output}
 
-        n_missing=\$(comm -3 <(zcat --force $scorefile | tail -n +2 | cut -f 1 | sort) <(sort ${output}.sscore.vars) | wc -l | tr -d ' ')
-
-        if [ \$n_missing -gt 0 ]
-        then
-          echo "ERROR: \$n_missing variant(s) missing from final calculated score!"
-          exit 1
-        else
-          echo "INFO: Scoring file variants match listed variants in sscore.vars"
-        fi
-
         cat <<-END_VERSIONS > versions.yml
         ${task.process.tokenize(':').last()}:
             plink2: \$(plink2 --version 2>&1 | sed 's/^PLINK v//; s/ 64.*\$//' )
@@ -93,16 +83,6 @@ process PLINK2_SCORE {
             $input ${geno.baseName} \
             --out ${output}
 
-        n_missing=\$(comm -3 <(zcat --force $scorefile | tail -n +2 | cut -f 1 | sort) <(sort ${output}.sscore.vars) | wc -l | tr -d ' ')
-
-        if [ \$n_missing -gt 0 ]
-        then
-          echo "ERROR: \$n_missing variant(s) missing from final calculated score!"
-          exit 1
-        else
-          echo "INFO: Scoring file variants match listed variants in sscore.vars"
-        fi
-
         cat <<-END_VERSIONS > versions.yml
         ${task.process.tokenize(':').last()}:
             plink2: \$(plink2 --version 2>&1 | sed 's/^PLINK v//; s/ 64.*\$//' )

diff --git a/modules/local/score_aggregate.nf b/modules/local/score_aggregate.nf
@@ -12,7 +12,9 @@ process SCORE_AGGREGATE {
         "${task.ext.docker}${task.ext.docker_version}" }"
 
     input:
-    tuple val(meta), path(scorefiles)
+    tuple val(meta), path(scorefiles) // calculated polygenic scores
+    path(scorefile_vars) // PGS scoring file
+    path(scored_vars) // variants _actually used_ to calculate scores
 
     output:
     tuple val(scoremeta), path("aggregated_scores.txt.gz"), emit: scores
@@ -21,7 +23,9 @@ process SCORE_AGGREGATE {
     script:
     scoremeta = meta.subMap('id')
     """
-    pgscatalog-aggregate -s $scorefiles -o . -v --no-split
+    # variants are always verified, so that variants in the scoring files
+    # overlap perfectly with the scored variants
+    pgscatalog-aggregate -s $scorefiles -o . -v --no-split --verify_variants
 
     cat <<-END_VERSIONS > versions.yml
     ${task.process.tokenize(':').last()}:

diff --git a/subworkflows/local/apply_score.nf b/subworkflows/local/apply_score.nf
@@ -123,7 +123,20 @@ workflow APPLY_SCORE {
         .map { [ it.first().subMap("id"), it.tail().findAll { !(it instanceof LinkedHashMap) }]}
         .set { ch_scores }
 
-    SCORE_AGGREGATE ( ch_scores )
+    // 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
+    PLINK2_SCORE.out.vars_scored
+        .collect()
+        .set { ch_vars_scored }
+
+    ch_target_scorefile.flatMap { it.last() }
+        .filter(Path)
+        .collect()
+        .set{ ch_scorefile_verify }
+
+    SCORE_AGGREGATE ( ch_scores, ch_vars_scored, ch_scorefile_verify )
+
     ch_versions = ch_versions.mix(SCORE_AGGREGATE.out.versions)
 
     emit: