Updated pangolin task to accept vcf with caller == input

Updated main.nf to respect parameter params.skip_normalization
Added integration tests for lineage mode and reference generate mode

main.nf

@@ -260,8 +260,13 @@ workflow {
             PANGOLIN_LINEAGE(input_fastas)
         }
         else if (input_vcfs) {
-            VARIANT_NORMALIZATION(input_vcfs, reference)
-            normalized_vcfs = VARIANT_NORMALIZATION.out
+            if (! params.skip_normalization) {
+                VARIANT_NORMALIZATION(input_vcfs, reference)
+                normalized_vcfs = VARIANT_NORMALIZATION.out
+            }
+            else {
+                normalized_vcfs = input_vcfs
+            }
             VCF2FASTA(normalized_vcfs, reference)
             PANGOLIN_LINEAGE(VCF2FASTA.out)
         }

modules/00_prepare_annotation.nf

@@ -18,6 +18,11 @@ process BWA_INDEX {
         path("reference/sequences.fa"), emit: reference
         path("reference/sequences.fa.fai"), emit: fai
         path("reference/sequences.dict"), emit: gatk_dict
+        path("reference/sequences.fa.0123")
+        path("reference/sequences.fa.amb")
+        path("reference/sequences.fa.ann")
+        path("reference/sequences.fa.bwt.2bit.64")
+        path("reference/sequences.fa.pac")
 
     script:
         memory = "${params.memory}".replaceAll(" ", "").toLowerCase()

modules/07_lineage_annotation.nf

@@ -19,7 +19,8 @@ process PANGOLIN_LINEAGE {
 
     when:
         // only runs pangolin on LoFreq and the assembly results
-        caller == "lofreq" || caller == "assembly"
+        // JoHa: added input to run lineage mode for vcf input
+        caller == "lofreq" || caller == "assembly" || caller == "input"
 
     shell:
     """

tests/scripts/test_14.sh

@@ -0,0 +1,32 @@
+#!/bin/bash
+
+##################################################################################
+# Genome generate
+##################################################################################
+echo "Running CoVigator pipeline test 14"
+source bin/assert.sh
+output=tests/output/test14
+nextflow main.nf -profile conda --name test_data \
+	--output $output \
+	--reference_generate \
+    --reference reference/Sars_cov_2.ASM985889v3.dna.toplevel.fa  \
+    --gff reference/Sars_cov_2.ASM985889v3.101.gff3 \
+    --snpeff_organism Sars_cov_2
+
+# Test reference genome related output
+test -s $output/reference/sequences.fa.fai || { echo "Missing fasta index file!"; exit 1; }
+test -s $output/reference/sequences.fa.dict || { echo "Missing GATK dict file!"; exit 1; }
+
+# Test bwa index files are present
+test -s $output/reference/sequences.fa.0123 || { echo "Missing bwa 0123 index file!"; exit 1; }
+test -s $output/reference/sequences.fa.amb || { echo "Missing bwa amb index file!"; exit 1; }
+test -s $output/reference/sequences.fa.ann || { echo "Missing bwa ann index file!"; exit 1; }
+test -s $output/reference/sequences.fa.bwt.2bit.64 || { echo "Missing bwa bwt.2bit.64 index file!"; exit 1; }
+test -s $output/reference/sequences.fa.pac || { echo "Missing bwa pac index file!"; exit 1; }
+
+
+# Test snpEff output
+test -s $output/snpeff/snpEff.config || { echo "Missing snpEff config file!"; exit 1; }
+test -s $output/snpeff/Sars_cov_2/snpEffectPredictor.bin || { echo "Missing snpEff predictor bin file!"; exit 1; }
+test -s $output/snpeff/Sars_cov_2/sequences.fa || { echo "Missing snpEff reference genome!"; exit 1; }
+test -s $output/snpeff/Sars_cov_2/genes.gff || { echo "Missing snpEff reference annotation!"; exit 1; }

tests/scripts/test_15.sh

@@ -0,0 +1,21 @@
+#!/bin/bash
+
+##################################################################################
+# Lineage only mode
+##################################################################################
+
+echo "Running CoVigator pipeline test 15"
+source bin/assert.sh
+output=tests/output/test15
+nextflow main.nf -profile conda --name test_data \
+	--output $output \
+	--vcf tests/test_data/test_data.lofreq.vcf \
+    --lineage_mode
+
+test -s $output/test_data.input.fasta || { echo "Missing VCF2FASTA fasta file (lineage mode with vcf input)!"; exit 1; }
+test -s $output/test_data.input.pangolin.csv || { echo "Missing pangolin output file (lineage mode with vcf input)!"; exit 1; }
+assert_eq `wc -l $output/test_data.input.pangolin.csv` 2 "Wrong number of pangolin results"
+
+
+
+

tests/scripts/test_16.sh

@@ -0,0 +1,16 @@
+#!/bin/bash
+
+##################################################################################
+# Lineage only mode
+##################################################################################
+
+echo "Running CoVigator pipeline test 16"
+source bin/assert.sh
+output=tests/output/test15
+nextflow main.nf -profile conda --name test_data \
+	--output $output \
+	--fasta tests/test_data/test_data.fasta \
+    --lineage_mode
+
+test -s $output/test_data.assembly.pangolin.csv || { echo "Missing pangolin output file (lineage mode with vcf input)!"; exit 1; }
+assert_eq `wc -l $output/test_data.assembly.pangolin.csv` 2 "Wrong number of pangolin results"