Accommodate Mbovis reference

.gitignore

@@ -29,6 +29,7 @@ samplesheet.csv
 magma.sh
 .Rproj.user
 
-docs/_site
+docs/.quarto
 
-.DS_Store
+docs/_site
+resources/snpeff/data/mbovis_AF2122/snpEffectPredictor.bin

bin/rename_vcf_chrom.py

@@ -36,7 +36,6 @@ def errlog(x,ext=False):
         quit(1)
 
 def cmd_out(cmd,verbose=1):
-    cmd = "set -u pipefail; " + cmd
     if verbose==2:
         sys.stderr.write("\nRunning command:\n%s\n" % cmd)
         stderr = open("/dev/stderr","w")
@@ -56,19 +55,27 @@ def cmd_out(cmd,verbose=1):
 
 def main(args):
     generator = cmd_out("bcftools view " + args.vcf) if args.vcf else sys.stdin
-    convert = dict(zip(args.source,args.target))
-    for l in generator:
-        if l[0]=="#":
-            sys.stdout.write(l.strip()+"\n")
-        else:
-            row = l.strip().split()
-            row[0] = convert[row[0]]
-            sys.stdout.write("\t".join(row)+"\n")
+    convert = dict(zip(args.source, args.target))
+
+    # Open the output file for writing, or use stdout if not specified
+    outfile = open(args.outfile, 'w') if args.outfile else sys.stdout
+    try:
+        for l in generator:
+            if l[0] == "#":
+                outfile.write(l.strip() + "\n")
+            else:
+                row = l.strip().split()
+                row[0] = convert[row[0]]
+                outfile.write("\t".join(row) + "\n")
+    finally:
+        if args.outfile:
+            outfile.close()
 
 parser = argparse.ArgumentParser(description='tbprofiler script',formatter_class=argparse.ArgumentDefaultsHelpFormatter)
 parser.add_argument('--vcf',type=str,help='')
 parser.add_argument('--source',nargs="+",type=str,help='')
 parser.add_argument('--target',nargs="+",type=str,help='')
+parser.add_argument('--outfile', type=str, required=False, help='Output VCF file')
 parser.set_defaults(func=main)
 args = parser.parse_args()
 args.func(args)

conf/docker.config

@@ -55,6 +55,12 @@ process {
         container = "ghcr.io/torch-consortium/magma/magma-container-2:1.1.1"
     }
 
+    // withName:
+    // 'SNPEFF.*' {
+    //     container = "quay.io/biocontainers/snpeff:4.3.1t--2"
+    // }
+
+
 }
 
 

default_params.config

@@ -487,7 +487,7 @@ GATK_GENOTYPE_GVCFS {
 SNPEFF {
     results_dir = "${params.outdir}/vcf_files/cohort/raw_variant_files/annotated"
 
-    arguments = " -nostats -ud 100 Mycobacterium_tuberculosis_h37rv "
+    arguments = " -nostats -ud 100  mbovis_AF2122 "
 
     should_publish = true
 }

modules.json

@@ -1,177 +1,26 @@
 {
-  "name": "TORCH-consortium/magma",
-  "homePage": "https://github.com/TORCH-consortium/MAGMA",
-  "repos": {
-    "https://github.com/nf-core/modules.git": {
-      "modules": {
-        "nf-core": {
-          "bcftools/merge": {
-            "branch": "master",
-            "git_sha": "c9c3ef86c1892413b3c86fb38c4e39fd7288512f",
-            "installed_by": ["modules"]
-          },
-          "bcftools/view": {
-            "branch": "master",
-            "git_sha": "c9c3ef86c1892413b3c86fb38c4e39fd7288512f",
-            "installed_by": ["modules"]
-          },
-          "bwa/mem": {
-            "branch": "master",
-            "git_sha": "a29f18660f5e3748d44d6f716241e70c942c065d",
-            "installed_by": ["modules"]
-          },
-          "bwamem2/mem": {
-            "branch": "master",
-            "git_sha": "a29f18660f5e3748d44d6f716241e70c942c065d",
-            "installed_by": ["modules"]
-          },
-          "delly/call": {
-            "branch": "master",
-            "git_sha": "05954dab2ff481bcb999f24455da29a5828af08d",
-            "installed_by": ["modules"]
-          },
-          "fastqc": {
-            "branch": "master",
-            "git_sha": "08108058ea36a63f141c25c4e75f9f872a5b2296",
-            "installed_by": ["modules"]
-          },
-          "fastqutils/info": {
-            "branch": "master",
-            "git_sha": "92168db7097574f7407062f26e901f2d2fb96135",
-            "installed_by": ["modules"]
-          },
-          "gatk4/applybqsr": {
-            "branch": "master",
-            "git_sha": "8a7e6e269f4044d922ce84bbe5717a8c07de8834",
-            "installed_by": ["modules"]
-          },
-          "gatk4/applyvqsr": {
-            "branch": "master",
-            "git_sha": "05954dab2ff481bcb999f24455da29a5828af08d",
-            "installed_by": ["modules"]
-          },
-          "gatk4/baserecalibrator": {
-            "branch": "master",
-            "git_sha": "f63c350bc9dc0b0a934d14d3a85298002c15adc1",
-            "installed_by": ["modules"]
-          },
-          "gatk4/combinegvcfs": {
-            "branch": "master",
-            "git_sha": "05954dab2ff481bcb999f24455da29a5828af08d",
-            "installed_by": ["modules"]
-          },
-          "gatk4/genotypegvcfs": {
-            "branch": "master",
-            "git_sha": "81880787133db07d9b4c1febd152c090eb8325dc",
-            "installed_by": ["modules"]
-          },
-          "gatk4/haplotypecaller": {
-            "branch": "master",
-            "git_sha": "05954dab2ff481bcb999f24455da29a5828af08d",
-            "installed_by": ["modules"]
-          },
-          "gatk4/indexfeaturefile": {
-            "branch": "master",
-            "git_sha": "05954dab2ff481bcb999f24455da29a5828af08d",
-            "installed_by": ["modules"]
-          },
-          "gatk4/markduplicates": {
-            "branch": "master",
-            "git_sha": "1ec937ab3edc307bc0d79a2200d784e9f0868359",
-            "installed_by": ["modules"]
-          },
-          "gatk4/mergevcfs": {
-            "branch": "master",
-            "git_sha": "05954dab2ff481bcb999f24455da29a5828af08d",
-            "installed_by": ["modules"]
-          },
-          "gatk4/selectvariants": {
-            "branch": "master",
-            "git_sha": "05954dab2ff481bcb999f24455da29a5828af08d",
-            "installed_by": ["modules"]
-          },
-          "gatk4/variantrecalibrator": {
-            "branch": "master",
-            "git_sha": "81880787133db07d9b4c1febd152c090eb8325dc",
-            "installed_by": ["modules"]
-          },
-          "gatk4/variantstotable": {
-            "branch": "master",
-            "git_sha": "05954dab2ff481bcb999f24455da29a5828af08d",
-            "installed_by": ["modules"]
-          },
-          "iqtree": {
-            "branch": "master",
-            "git_sha": "2ae11ef01a83c7330c09d08d14d21ff445e32141",
-            "installed_by": ["modules"]
-          },
-          "ismapper": {
-            "branch": "master",
-            "git_sha": "81880787133db07d9b4c1febd152c090eb8325dc",
-            "installed_by": ["modules"]
-          },
-          "lofreq/call": {
-            "branch": "master",
-            "git_sha": "81880787133db07d9b4c1febd152c090eb8325dc",
-            "installed_by": ["modules"]
-          },
-          "lofreq/filter": {
-            "branch": "master",
-            "git_sha": "05954dab2ff481bcb999f24455da29a5828af08d",
-            "installed_by": ["modules"]
-          },
-          "lofreq/indelqual": {
-            "branch": "master",
-            "git_sha": "05954dab2ff481bcb999f24455da29a5828af08d",
-            "installed_by": ["modules"]
-          },
-          "multiqc": {
-            "branch": "master",
-            "git_sha": "f0719ae309075ae4a291533883847c3f7c441dad",
-            "installed_by": ["modules"]
-          },
-          "samtools/index": {
-            "branch": "master",
-            "git_sha": "d090922b1a4b80ba283186459ababf8e308abcbb",
-            "installed_by": ["modules"]
-          },
-          "samtools/merge": {
-            "branch": "master",
-            "git_sha": "d090922b1a4b80ba283186459ababf8e308abcbb",
-            "installed_by": ["modules"]
-          },
-          "samtools/stats": {
-            "branch": "master",
-            "git_sha": "05954dab2ff481bcb999f24455da29a5828af08d",
-            "installed_by": ["modules"]
-          },
-          "snpdists": {
-            "branch": "master",
-            "git_sha": "81880787133db07d9b4c1febd152c090eb8325dc",
-            "installed_by": ["modules"]
-          },
-          "snpeff/download": {
-            "branch": "master",
-            "git_sha": "81880787133db07d9b4c1febd152c090eb8325dc",
-            "installed_by": ["modules"]
-          },
-          "snpeff/snpeff": {
-            "branch": "master",
-            "git_sha": "05954dab2ff481bcb999f24455da29a5828af08d",
-            "installed_by": ["modules"]
-          },
-          "snpsites": {
-            "branch": "master",
-            "git_sha": "05954dab2ff481bcb999f24455da29a5828af08d",
-            "installed_by": ["modules"]
-          },
-          "tbprofiler/profile": {
-            "branch": "master",
-            "git_sha": "81880787133db07d9b4c1febd152c090eb8325dc",
-            "installed_by": ["modules"]
-          }
+    "name": "TORCH-consortium/magma",
+    "homePage": "https://github.com/TORCH-consortium/MAGMA",
+    "repos": {
+        "https://github.com/nf-core/modules.git": {
+            "modules": {
+                "nf-core": {
+                    "snpeff/download": {
+                        "branch": "master",
+                        "git_sha": "81880787133db07d9b4c1febd152c090eb8325dc",
+                        "installed_by": [
+                            "modules"
+                        ]
+                    },
+                    "snpeff/snpeff": {
+                        "branch": "master",
+                        "git_sha": "05954dab2ff481bcb999f24455da29a5828af08d",
+                        "installed_by": [
+                            "modules"
+                        ]
+                    }
+                }
+            }
         }
-      }
     }
-  }
-}
+}

modules/local/gatk/variant_recalibrator.nf

@@ -27,7 +27,6 @@ process GATK_VARIANT_RECALIBRATOR {
     tag "annotation: ${annotations}"
     publishDir params.results_dir, mode: params.save_mode, enabled: params.should_publish
 
-
     input:
         val(analysisMode)
         val(annotations)
@@ -47,16 +46,11 @@ process GATK_VARIANT_RECALIBRATOR {
         tuple val(joint_name), path("*${analysisMode}*.command.log"), emit: annotationsLog
 
     script:
-
-        def finalResourceFilesArg =    (resourceFilesArg  ? "--resource:${resourceFilesArg}" : "")
-
-        def optionalAnnotationPrefix = ""
-
-        if (task.process.split("__").length == 1) {
-            optionalAnnotationPrefix = ""
-        } else {
-            optionalAnnotationPrefix = ".${task.process.split("__")[-1]}"
-        }
+        // Only add --resource argument if resourceFiles and resourceFileIndexes are non-empty
+        def resourceFilesExist = resourceFiles && resourceFiles.size() > 0
+        def resourceFileIndexesExist = resourceFileIndexes && resourceFileIndexes.size() > 0
+        def finalResourceFilesArg = (resourceFilesArg && resourceFilesExist && resourceFileIndexesExist) ? "--resource:${resourceFilesArg}" : ""
+        def optionalAnnotationPrefix = (task.process.split("__").length == 1) ? "" : ".${task.process.split('__')[-1]}"
 
         """
         ${params.gatk_path} VariantRecalibrator --java-options "-Xmx${task.memory.giga}G" \\
@@ -73,15 +67,5 @@ process GATK_VARIANT_RECALIBRATOR {
             2>${joint_name}.${analysisMode}${optionalAnnotationPrefix}.command.log
 
         cp ${joint_name}.${analysisMode}${optionalAnnotationPrefix}.command.log .command.log
-
-        """
-
-    stub:
-
-        """
-        touch ${joint_name}.${analysisMode}.tranches
-        touch ${joint_name}.${analysisMode}.R
-        touch ${joint_name}.${analysisMode}.recal.vcf.gz
-        touch ${joint_name}.${analysisMode}.mod
         """
 }

modules/local/snpeff/snpeff.nf

@@ -30,6 +30,8 @@ process SNPEFF {
     input:
         tuple val(joint_name), path(rawJointVariantsFile)
         path(ref_fasta)
+        path(snpeff_config)
+        path(snpeff_db)
 
     output:
         tuple val(joint_name), path("*.annotated.vcf")
@@ -38,9 +40,11 @@ process SNPEFF {
     shell:
 
         '''
-        rename_vcf_chrom.py --vcf !{rawJointVariantsFile}  --source !{params.ref_fasta_basename} --target 'Chromosome' \
-            | !{params.snpeff_path} -nostats !{params.arguments}  \
-            | rename_vcf_chrom.py --target !{params.ref_fasta_basename} --source 'Chromosome' \
+        snpEff build -c complete_snpeff.config -gff3 -v mbovis_AF2122
+
+        rename_vcf_chrom.py --vcf !{rawJointVariantsFile}  --source !{params.ref_fasta_basename} --target 'Chromosome' \\
+            | !{params.snpeff_path} -nostats -c !{snpeff_config} !{params.arguments}  \\
+            | rename_vcf_chrom.py --target !{params.ref_fasta_basename} --source 'Chromosome' \\
          > !{joint_name}.raw_variants.annotated.vcf
         '''
 

modules/local/tbprofiler/vcf_profile__cohort.nf

@@ -40,8 +40,8 @@ process TBPROFILER_VCF_PROFILE__COHORT {
 
         """
 
-        bcftools view ${mergedVcf} | sed 's/NC-000962-3-H37Rv/Chromosome/g' > intermediate.vcf
-         
+        bcftools view ${mergedVcf} | sed 's/${params.ref_fasta_basename}/Chromosome/g' > intermediate.vcf
+
         cat  intermediate.vcf | bcftools view -Oz -o intermediate.vcf.gz
 
         ${params.tbprofiler_path} profile  \\

modules/local/tbprofiler/vcf_profile__lofreq.nf

@@ -41,8 +41,8 @@ process TBPROFILER_VCF_PROFILE__LOFREQ {
 
         """
 
-        bcftools view ${mergedLofreqVcf} | sed 's/NC-000962-3-H37Rv/Chromosome/g' > intermediate.vcf
-         
+        bcftools view ${mergedLofreqVcf} | sed 's/${params.ref_fasta_basename}/Chromosome/g' > intermediate.vcf
+
         cat  intermediate.vcf | bcftools view -Oz -o intermediate.vcf.gz
 
         ${params.tbprofiler_path} profile \\