Merge branch 'main' into features/pgs-calculation

genepi · Nov 21, 2023 · 1c0081c · 1c0081c
2 parents 9ea2973 + 594c40a
commit 1c0081c
Show file tree

Hide file tree

Showing 8 changed files with 68 additions and 21 deletions.
diff --git a/Dockerfile b/Dockerfile
@@ -1,11 +1,11 @@
-FROM ubuntu:18.04
+FROM ubuntu:22.04
 MAINTAINER Lukas Forer <[email protected]> / Sebastian Schönherr <[email protected]>
 
 # Install compilers
 RUN apt-get update && apt-get install -y wget build-essential zlib1g-dev liblzma-dev libbz2-dev libxau-dev
 
 #  Install miniconda
-RUN wget --quiet https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda.sh && \
+RUN wget --quiet https://repo.anaconda.com/miniconda/Miniconda3-py39_23.9.0-0-Linux-x86_64.sh -O ~/miniconda.sh && \
   /bin/bash ~/miniconda.sh -b -p /opt/conda
 ENV PATH=/opt/conda/bin:${PATH}
 
@@ -79,5 +79,3 @@ ENV JAVA_TOOL_OPTIONS="-Djdk.lang.Process.launchMechanism=vfork"
 
 COPY files/bin/trace /usr/bin/.
 COPY files/bin/vcf2geno /usr/bin/.
-
-
diff --git a/modules/local/compression/compression_encryption_vcf.nf b/modules/local/compression/compression_encryption_vcf.nf
@@ -18,7 +18,7 @@ process COMPRESSION_ENCRYPTION_VCF {
     def info_joined = ArrayUtil.sort(imputed_info)
     def prefix = "chr${chr}"
     def imputed_name = "${prefix}.dose.vcf.gz"
-    def meta_name = "${prefix}_empiricalDose.vcf.gz"
+    def meta_name = "${prefix}.empiricalDose.vcf.gz"
     def zip_name = "chr_${chr}.zip"
     def info_name = "${prefix}.info"
     def aes = params.encryption.aes ? "-mem=AES256" : ""

diff --git a/modules/local/phasing/beagle.nf b/modules/local/phasing/beagle.nf
@@ -3,7 +3,7 @@ process BEAGLE {
     tag "${chunkfile}"
 
     input:
-    tuple val(chr), path(bcf), val(start), val(end), val(phasing_status), path(chunkfile), val(snps), val(in_reference),  path(map_beagle)
+    tuple val(chr), path(bcf), val(start), val(end), val(phasing_status), path(chunkfile), path(map_beagle)
     val phasing_method
 
     output:

diff --git a/modules/local/phasing/eagle.nf b/modules/local/phasing/eagle.nf
@@ -3,7 +3,7 @@ process EAGLE {
     tag "${chunkfile}"
 
     input:
-    tuple val(chr), path(bcf), path(bcf_csi), val(start), val(end), val(phasing_status), path(chunkfile), val(snps), val(in_reference)
+    tuple val(chr), path(bcf), path(bcf_csi), val(start), val(end), val(phasing_status), path(chunkfile)
     path map_eagle
 
     output:

diff --git a/nf-test.config b/nf-test.config
@@ -5,4 +5,8 @@ config {
     configFile "tests/nextflow.config"
     profile "development"
 
+   plugins {
+     load "[email protected]"
+  }
+
 }
diff --git a/tests/main.nf.test b/tests/main.nf.test
@@ -66,18 +66,19 @@ nextflow_pipeline {
         }
 
     }
-
+    
     test("Should run with phased data") {
 
         when {
             params {
-                project  = "test-job"
+                project  = "testPipelineWithPhased"
                 build = "hg19"
                 files = "$projectDir/tests/input/chr20-phased/*.vcf.gz"
                 population = "eur"
                 password = PASSWORD
                 refpanel_yaml = "$projectDir/tests/data/refpanels/hapmap2-chr20/cloudgene.yaml"
                 output = "${outputDir}"
+                phasing = "no_phasing"
             }
         }
 
@@ -92,19 +93,56 @@ nextflow_pipeline {
             assert imputed_chr_20.exists()
             ZipFile zipFile = new ZipFile(imputed_chr_20, PASSWORD.toCharArray());
             zipFile.extractAll("${outputDir}");
+            def file = path("${outputDir}/chr20.dose.vcf.gz").vcf
+            assert file.getChromosome() == "20"
+            assert file.getNoSamples() == 51;
+            assert file.isPhased()
+            assert file.getNoSnps() == TOTAL_REFPANEL_CHR20_B37 + ONLY_IN_INPUT
+
+            //check correct number of snps in info.gz file
+            assert path("${outputDir}/chr20.info.gz").linesGzip.size() == 1 + TOTAL_REFPANEL_CHR20_B37 + ONLY_IN_INPUT
+
+        }
+
+    }
+
+    test("Should run with phased data and meta option") {
+
+        when {
+            params {
+                project  = "testPipelineWithPhasedAndMetaOption"
+                build = "hg19"
+                files = "$projectDir/tests/input/chr20-phased/*.vcf.gz"
+                population = "eur"
+                password = PASSWORD
+                refpanel_yaml = "$projectDir/tests/data/refpanels/hapmap2-chr20/cloudgene.yaml"
+                output = "${outputDir}"
+                phasing = "no_phasing"
+            }
+        }
 
-            //TODO: this is from imputationserver-utils.jar. How to include? use alternative?
-            /*VcfFile file = VcfFileUtil.load("${outputDir}/chr20.dose.vcf.gz", 100000000, false);
+        then {
+            assert workflow.success
 
+            def quality_control_log = file("${outputDir}/cloudgene.report.json")
+            assert quality_control_log.exists()
+            assert quality_control_log.text.contains("Remaining sites in total: 7,735")
+
+            def imputed_chr_20 = file("${outputDir}/chr_20.zip");
+            assert imputed_chr_20.exists()
+            ZipFile zipFile = new ZipFile(imputed_chr_20, PASSWORD.toCharArray());
+            zipFile.extractAll("${outputDir}");
+            def file = path("${outputDir}/chr20.dose.vcf.gz").vcf
+            def fileMeta = path("${outputDir}/chr20.empiricalDose.vcf.gz").vcf
+            assert fileMeta.getNoSnps() == 7735
             assert file.getChromosome() == "20"
             assert file.getNoSamples() == 51;
             assert file.isPhased()
-            assert file.getNoSnps() == TOTAL_REFPANEL_CHR20_B37 + ONLY_IN_INPUT*/
-
+            assert file.getNoSnps() == TOTAL_REFPANEL_CHR20_B37 + ONLY_IN_INPUT
+            
             //check correct number of snps in info.gz file
-            //TODO: phasing executed every time, therefore ONLY_IN_INPUT is currently not available
-           // assert path("${outputDir}/chr20.info.gz").linesGzip.size() == 1 + TOTAL_REFPANEL_CHR20_B37 + ONLY_IN_INPUT
-            assert path("${outputDir}/chr20.info.gz").linesGzip.size() == 1 + TOTAL_REFPANEL_CHR20_B37
+            assert path("${outputDir}/chr20.info.gz").linesGzip.size() == 1 + TOTAL_REFPANEL_CHR20_B37 + ONLY_IN_INPUT
+
         }
 
     }

diff --git a/workflows/imputationserver.nf b/workflows/imputationserver.nf
@@ -53,17 +53,24 @@ workflow IMPUTATIONSERVER {
             INPUT_VALIDATION.out,
             legend_files_ch.collect()
         )
-        //TODO: add phasing only mode
+
         if (params.mode != 'qc-only') {
 
-            PHASING(
-                QUALITY_CONTROL.out.qc_metafiles
+            imputation_ch =  QUALITY_CONTROL.out.qc_metafiles
+
+            if ("${params.phasing}" != 'no_phasing') { 
+
+                PHASING(
+                    imputation_ch
             )
 
             if (params.mode == 'imputation') {
+            imputation_ch = PHASING.out.phased_ch
+
+            }
 
                 IMPUTATION(
-                    PHASING.out.phased_ch
+                    imputation_ch
                 )
 
                 ENCRYPTION(

diff --git a/workflows/quality_control.nf b/workflows/quality_control.nf
@@ -29,7 +29,7 @@ workflow QUALITY_CONTROL {
     chunks_csv_index
         .combine(chunks_vcf_index, by: 0)
         .map{
-            row-> tuple(row[1], row[2], row[3], row[4], file(row[8]), row[6], row[7])
+            row-> tuple(row[1], row[2], row[3], row[4], file(row[8]))
         }
         .set { metafiles_ch }