add PCA (to be renamed) subworkflow to the targene workflow

TARGENE · Jun 5, 2024 · 58a31e1 · 58a31e1
1 parent a204a32
commit 58a31e1
Show file tree

Hide file tree

Showing 4 changed files with 20 additions and 50 deletions.
diff --git a/modules/simulations.nf b/modules/simulations.nf
@@ -33,7 +33,7 @@ process NullSimulationEstimation {
 
 process RealisticSimulationEstimation {
     label 'simulation_image'
-    publishDir "${params.OUTDIR}/null_simulation_results", mode: 'symlink'
+    publishDir "${params.OUTDIR}/realistic_simulation_results", mode: 'symlink'
 
     input:
         path origin_dataset

diff --git a/workflows/pca.nf b/workflows/pca.nf
@@ -34,5 +34,8 @@ workflow PCA {
     GeneticConfounders(IIDGenotypes.out)
 
     emit:
-        GeneticConfounders.out
+        traits = ExtractTraits.out
+        iid_genotypes = IIDGenotypes.out
+        pcs = GeneticConfounders.out
+
 }
diff --git a/workflows/simulations.nf b/workflows/simulations.nf
@@ -18,17 +18,14 @@ workflow NULL_SIMULATION {
     rngs = Channel.fromList(params.RNGS)
     bgen_files = Channel.fromPath("$params.BGEN_FILES", checkIfExists: true).collect()
 
-    // Extract Traits
-    traits = EXTRACT_TRAITS()
-
     // PCA
-    pcs = PCA()
+    PCA()
 
     // generate main dataset and estimand configuration files
     EstimationInputs(
         bgen_files,
-        traits,
-        pcs,
+        PCA.out.traits,
+        PCA.out.pcs,
         estimands_files,
         bqtls_file,
         transactors_files,
@@ -62,19 +59,16 @@ workflow REALISTIC_SIMULATION {
     sample_sizes = Channel.fromList(params.SAMPLE_SIZES)
     rngs = Channel.fromList(params.RNGS)
     bgen_files = Channel.fromPath("$params.BGEN_FILES", checkIfExists: true).collect()
-
-    // Extract Traits
-    traits = EXTRACT_TRAITS()
 
     // PCA
-    pcs = PCA()
+    PCA()
 
     // Realistic Simulation Inputs
     simulation_inputs = RealisticSimulationInputs(
         estimands_files.collect(),
         bgen_files,
-        traits,
-        pcs,
+        PCA.out.traits,
+        PCA.out.pcs,
         ga_trait_table
     )
     dataset = simulation_inputs.dataset

diff --git a/workflows/targene.nf b/workflows/targene.nf
@@ -1,56 +1,29 @@
+include { PCA } from './pca.nf'
 include { EstimationInputs } from '../subworkflows/estimation_inputs.nf'
-include { IIDGenotypes; GeneticConfounders } from '../subworkflows/confounders.nf'
-include { ExtractTraits } from '../subworkflows/extract_traits.nf'
 include { EstimationWorkflow } from '../subworkflows/estimation.nf'
 include { SVPWorkflow } from '../subworkflows/svp.nf'
 
 workflow TARGENE {
     // Define Parameters
     bgen_files = Channel.fromPath("$params.BGEN_FILES", checkIfExists: true).collect()
     estimands_file = Channel.value(file("$params.ESTIMANDS_FILE"))
+    estimator_config = Channel.value(file("${params.ESTIMATOR_FILE}"))
+
+    // DEPRECATED: LEGACY TO BE REMOVED
     bqtls_file = Channel.value(file("$params.BQTLS"))
     transactors_files = Channel.fromPath("$params.TRANS_ACTORS").collect()
     extra_confounders = Channel.value(file("$params.EXTRA_CONFOUNDERS"))
     extra_treatments = Channel.value(file("$params.ENVIRONMENTALS"))
     extra_covariates = Channel.value(file("$params.EXTRA_COVARIATES"))
 
-    ukb_encoding_file = params.UKB_ENCODING_FILE
-    ukb_config = Channel.value(file("$params.UKB_CONFIG", checkIfExists: true))
-    ukb_withdrawal_list = Channel.value(file("$params.UKB_WITHDRAWAL_LIST", checkIfExists: true))
-    traits_dataset = Channel.value(file("$params.TRAITS_DATASET", checkIfExists: true))
-
-    qc_file = Channel.value(file("$params.QC_FILE", checkIfExists: true))
-    flashpca_excl_reg = Channel.value(file("$params.FLASHPCA_EXCLUSION_REGIONS", checkIfExists: true))
-    ld_blocks = Channel.value(file("$params.LD_BLOCKS", checkIfExists: true))
-    bed_files = Channel.fromFilePairs("$params.BED_FILES", size: 3, checkIfExists: true){ file -> file.baseName }
-
-    estimator_config = Channel.value(file("${params.ESTIMATOR_FILE}"))
-
     // Extract Traits
-    ExtractTraits(
-        traits_dataset,
-        ukb_config,
-        ukb_withdrawal_list,
-        ukb_encoding_file,
-    )
-
-    // IID Genotypes
-    IIDGenotypes(
-        flashpca_excl_reg,
-        ld_blocks,
-        bed_files,
-        qc_file,
-        ExtractTraits.out,
-    )
-
-    // Genetic confounders
-    GeneticConfounders(IIDGenotypes.out)
+    PCA()
 
     // generate main dataset and estimand configuration files
     EstimationInputs(
         bgen_files,
-        ExtractTraits.out,
-        GeneticConfounders.out,
+        PCA.out.traits,
+        PCA.out.pcs,
         estimands_file,
         bqtls_file,
         transactors_files,
@@ -66,11 +39,11 @@ workflow TARGENE {
         estimator_config,
     )
 
-    // Generate sieve estimates
+    // Generate sieve variance plateau estimates
     if (params.SVP == true){
         sieve_results = SVPWorkflow(
             EstimationWorkflow.out.hdf5_result, 
-            IIDGenotypes.out,
+            PCA.out.iid_genotypes,
         )
     }
 }