Poc/straglr

config/nxf_call_str.config

@@ -2,7 +2,7 @@ includeConfig 'nxf.config'
 
 env {
   CMD_EXPANSIONHUNTER = "apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/expansionhunter-5.0.0.sif ExpansionHunter"
-  CMD_STRAGLR="apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/straglr-1.4.4_vip_v3.sif straglr-genotype"
+  CMD_STRAGLR="apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/straglr-1.5.1.sif straglr.py"
 }
 
 params {

config/nxf_cram.config

@@ -6,7 +6,7 @@ includeConfig 'nxf_call_sv.config'
 
 params {
   cram {
-    call_snv = true
+    call_snv = false
     call_str = true
     call_sv = true
     call_cnv = true

config/nxf_vcf.config

@@ -8,7 +8,7 @@ env {
   CMD_FILTERVEP = "apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/vep-111.0.sif filter_vep"
   CMD_STRANGER = "apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/stranger-0.8.1.sif stranger"
   CMD_VCFREPORT="apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/vcf-report-7.0.0.sif"
-  CMD_VCFDECISIONTREE = "apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/vcf-decision-tree-4.1.1.sif"
+  CMD_VCFDECISIONTREE = "apptainer exec --no-mount home --bind \${TMPDIR} ${projectDir}/_dev/images/vcf-decision-tree-4.2.0-beta.sif"
   CMD_VCFINHERITANCEMATCHER = "apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/vcf-inheritance-matcher-3.1.0.sif"
 
   // workaround for SAMtools https://github.com/samtools/samtools/issues/1366#issuecomment-769170935
@@ -33,7 +33,7 @@ process {
   }
 
   withLabel:'vcf_classify|vcf_classify_samples|vcf_inheritance' {
-    memory = '2GB'
+    memory = '4GB'
   }
 
   withLabel: 'vcf_report' {
@@ -64,7 +64,8 @@ params {
       GRCh38 {
         capice_model = "${projectDir}/resources/GRCh38/capice_model_v5.1.2-v2.ubj"
         expansionhunter_variant_catalog = "${projectDir}/resources/GRCh38/expansionhunter_variant_catalog.json"
-        stranger_catalog = "${projectDir}/resources/GRCh38/variant_catalog_grch38_fixed.json"
+        str_catalog = "${projectDir}/resources/GRCh38/str_catalog_v1.beta.txt"
+        str_header = "${projectDir}/resources/GRCh38/str_header.hdr"
         vep_custom_phylop = "${projectDir}/resources/GRCh38/hg38.phyloP100way.bw"
         vep_plugin_clinvar = "${projectDir}/resources/GRCh38/clinvar_20240603_stripped.tsv.gz"
         vep_plugin_gnomad = "${projectDir}/resources/GRCh38/gnomad.total.v4.1.sites.stripped.tsv.gz"
@@ -99,19 +100,19 @@ params {
     }
 
     filter {
-      classes = "VUS,LP,P"
+      classes = "VUS,LP,P,s"
       consequences = true
     }
 
     filter_samples {
-      classes = "U1,U2"
+      classes = "U1,U2,LB,VUS,LP"
     }
 
     report {
       include_crams = true
       max_records = ""
       max_samples = ""
-      template = "${projectDir}/resources/vip-report-template-v6.2.0.html"
+      template = "${projectDir}/resources/vip-report-template-v6.2.0-str-poc.html"
 			metadata = "${projectDir}/resources/field_metadata.json"
 
       GRCh38 {

install.sh

@@ -83,11 +83,9 @@ download_files() {
   urls+=("0efcb85f297f08486cd01690b5f13ba0" "images/mosdepth-0.3.8.sif")
   urls+=("06ac8a76a307fa42fffd80ab906fd24b" "images/picard-3.1.1.sif")
   urls+=("9a4b685b26744113d3ea0a3904c02706" "images/samtools-1.17-patch1.sif")
-  urls+=("33f84edc86db09103d835748905fca25" "images/seqtk-1.4.sif")
-  urls+=("f4a3bef9cf6c6bc63c454c211e692b31" "images/spectre-0.2.1-patched.sif")
-  urls+=("8f6e06847776448e004df8b863571109" "images/straglr-1.4.4_vip_v3.sif")
+  urls+=("14d9cc4a5b3c6d705eb37cc1aeead9e8" "images/straglr-1.5.1.sif")
   urls+=("bcc157242cd9b09c66f015c52ef2d61d" "images/stranger-0.8.1.sif")
-  urls+=("57401e7b835fed2f52fafadc0dd744d4" "images/vcf-decision-tree-4.1.1.sif")
+  urls+=("177bc349312820641800fced563bddb4" "_dev/images/vcf-decision-tree-4.2.0-beta.sif")
   urls+=("9c4d7b48138f29651cdd45eb8d0cc4b6" "images/vcf-inheritance-matcher-3.1.0.sif")
   urls+=("53f9265acb2041b2b93c692177d91d74" "images/vcf-report-7.0.0.sif")
   urls+=("7bffc236a7c65b2b2e2e5f7d64beaa87" "images/vep-111.0.sif")
@@ -146,7 +144,7 @@ download_files() {
   # update utils/install.sh when updating inheritance.tsv
   urls+=("df31eb0fe9ebd9ae26c8d6f5f7ba6e57" "resources/inheritance_20240115.tsv")
   urls+=("7138e76a38d6f67935699d06082ecacf" "resources/vep/cache/homo_sapiens_refseq_vep_111_GRCh38.tar.gz")
-  urls+=("78962f0c7c6fe5c63ef7c66b627c95a0" "resources/vip-report-template-v6.2.0.html")
+  urls+=("0bd1d5f66f4cf90038c73c1ba72dd820" "resources/vip-report-template-v6.2.0-str-poc.html")
   # when modifying urls array, please keep list in 'ls -l' order
   for ((i = 0; i < ${#urls[@]}; i += 2)); do
     download_file "${base_url}" "${urls[i+1]}" "${urls[i+0]}" "${output_dir}" "${validate}"

modules/cram/templates/straglr_call.sh

@@ -5,27 +5,31 @@ call_short_tandem_repeats () {
     local args=()
     args+=("--loci" "!{paramLoci}")
     args+=("--sample" "!{sampleId}")
-    args+=("--vcf" "straglr.vcf")
-    args+=("--tsv" "!{tsvOut}")
     if [ -z "!{sampleSex}" ]; then
         args+=("--sex" "!{sampleSex}")
     fi
     args+=("--min_support" "!{paramMinSupport}")
     args+=("--min_cluster_size" "!{paramMinClusterSize}")
     args+=("!{cram}")
     args+=("!{paramReference}")
+    args+=("straglr")
 
     ${CMD_STRAGLR} "${args[@]}"
+
+    mv straglr.tsv "!{tsvOut}"
 }
 
 index () {
-  # workaround for https://github.com/molgenis/vip/issues/471
-  ${CMD_BCFTOOLS} reheader --fai "!{paramReferenceFai}" --temp-prefix . --threads "!{task.cpus}" "straglr.vcf" | ${CMD_BCFTOOLS} sort --temp-dir . --max-mem "!{task.memory.toGiga() - 1}G" --output-type z --output "!{vcfOut}"
+  awk '/#CHROM*/{print "##INFO=<ID=SVTYPE,Number=1,Type=String,Description=\"Type of structural variant\">"}1' ./straglr.vcf |\
+  awk 'BEGIN{FS=OFS="\t"} /^#/ {print; next} {$8="SVTYPE=STR;"$8; print; next;} { print; }'h > straglr_headered.vcf
+  ${CMD_BCFTOOLS} reheader --fai "!{paramReferenceFai}" --temp-prefix . --threads "!{task.cpus}" straglr_headered.vcf |\
+  ${CMD_BCFTOOLS} sort --temp-dir . --max-mem "!{task.memory.toGiga() - 1}G" --output-type z --output "!{vcfOut}"
 
   ${CMD_BCFTOOLS} index --csi --output "!{vcfOutIndex}" --threads "!{task.cpus}" "!{vcfOut}"
   ${CMD_BCFTOOLS} index --stats "!{vcfOut}" > "!{vcfOutStats}"
 
   rm straglr.vcf
+  rm straglr_headered.vcf
 }
 
 main() {

modules/vcf/annotate.nf

@@ -33,7 +33,8 @@ process annotate {
     reMMScoresPath = params.vcf.annotate[assembly].vep_plugin_ReMM_scores
     capiceModelPath = params.vcf.annotate[assembly].capice_model
     alphScorePath = params.vcf.annotate[assembly].vep_plugin_alphscore
-    strangerCatalog = params.vcf.annotate[assembly].stranger_catalog
+    strCatalog = params.vcf.annotate[assembly].str_catalog
+    strHeader = params.vcf.annotate[assembly].str_header
 
     areProbandHpoIdsIndentical = areProbandHpoIdsIndentical(meta.project.samples)
     gadoScores = meta.gado != null ? meta.gado : ""

modules/vcf/templates/annotate.sh

@@ -114,15 +114,13 @@ capice_predict() {
 }
 
 stranger() {
-    cp "!{vcfOut}" stranger_input.vcf.gz
-
-    local args=()
-    args+=("--repeats-file" "!{strangerCatalog}")
-    args+=("--loglevel" "ERROR")
-    args+=("stranger_input.vcf.gz")
-
-    ${CMD_STRANGER} "${args[@]}" | ${CMD_BCFTOOLS} view --no-version --threads "!{task.cpus}" --output-type "z" --output-file "!{vcfOut}"
-    rm "stranger_input.vcf.gz"
+    #FIXME: annotate strips all but STR's due to -i
+    #e.g. strip out all non-STR to separate file and merge after annotation
+    cp "!{vcfOut}" str_input.vcf.gz
+    zcat str_input.vcf.gz |\
+    ${CMD_BCFTOOLS} annotate -a "!{strCatalog}" -c CHROM,FROM,TO,repeat_unit,Repeat_id,Gene_id,Disease,HGNCId,-,-,LocusStructure,NormalMax,PathogenicMin -i 'SVTYPE="STR"' -h "!{strHeader}" |\
+    ${CMD_BCFTOOLS} view --no-version --threads "!{task.cpus}" --output-type "z" --output-file "!{vcfOut}"
+    rm "str_input.vcf.gz"
 }
 
 vep() {
@@ -215,7 +213,7 @@ main () {
   fi
   capice
   vep
-  if [ -n "!{strangerCatalog}" ]; then
+  if [ -n "!{strCatalog}" ]; then
     stranger
   fi
   index

resources/decision_tree_GRCh38.json

@@ -168,7 +168,7 @@
         "nextNode": "gnomAD"
       },
       "outcomeFalse": {
-        "nextNode": "exit_rm"
+        "nextNode": "gnomAD"
       }
     },
     "sv": {
@@ -191,7 +191,7 @@
         "value": "STR"
       },
       "outcomeTrue": {
-        "nextNode": "str_status"
+        "nextNode": "str_sequence"
       },
       "outcomeFalse": {
         "nextNode": "annotSV"
@@ -200,25 +200,21 @@
         "nextNode": "annotSV"
       }
     },
-    "str_status": {
-      "description": "Stranger str status (normal, pre_mutation, mutation)",
-      "type": "CATEGORICAL",
-      "field": "INFO/STR_STATUS",
-      "outcomeMap": {
-        "full_mutation": {
-          "nextNode": "exit_lp"
-        },
-        "pre_mutation": {
-          "nextNode": "exit_vus"
-        },
-        "normal": {
-          "nextNode": "exit_lb"
-        }
+    "str_sequence": {
+      "description": "STR sequence matches requested repeat unit",
+      "type": "BOOL",
+      "query": {
+        "field": "INFO/repeat_unit",
+        "operator": "eq_seq",
+        "value": "field:INFO/RU"
+      },
+      "outcomeTrue": {
+        "nextNode": "exit_str"
       },
       "outcomeMissing": {
         "nextNode": "exit_vus"
       },
-      "outcomeDefault": {
+      "outcomeFalse": {
         "nextNode": "exit_vus"
       }
     },
@@ -430,6 +426,11 @@
       "description": "Pathogenic",
       "type": "LEAF",
       "class": "P"
+    },
+    "exit_str": {
+      "description": "Short Tandem Repeat - pathogenicity based on sample info",
+      "type": "LEAF",
+      "class": "s"
     }
   }
 }

resources/decision_tree_samples.json

@@ -1,6 +1,24 @@
 {
-  "rootNode": "gt",
+  "rootNode": "vipc",
   "nodes": {
+    "vipc": {
+      "type": "BOOL",
+      "description": "Genotype quality",
+      "query": {
+        "field": "INFO/CSQ/VIPC",
+        "operator": "==",
+        "value": "s"
+      },
+      "outcomeTrue": {
+        "nextNode": "str_pathogenic_min"
+      },
+      "outcomeFalse": {
+        "nextNode": "gt"
+      },
+      "outcomeMissing": {
+        "nextNode": "gt"
+      }
+    },
     "gt": {
       "description": "Genotype",
       "type": "CATEGORICAL",
@@ -86,6 +104,42 @@
         "nextNode": "exit_u2"
       }
     },
+    "str_pathogenic_min": {
+      "description": "Nr of repeat units above pathogenic_min",
+      "type": "BOOL",
+      "query": {
+        "field": "FORMAT/AC",
+        "operator": "range_above",
+        "value": "field:INFO/PathogenicMin"
+      },
+      "outcomeTrue": {
+        "nextNode": "str_pathogenic_min"
+      },
+      "outcomeMissing": {
+        "nextNode": "exit_vus"
+      },
+      "outcomeFalse": {
+        "nextNode": "exit_vus"
+      }
+    },
+    "str_normal_max": {
+      "description": "Range of repeat units below normal_max",
+      "type": "BOOL",
+      "query": {
+        "field": "FORMAT/ACR",
+        "operator": "range_below",
+        "value": "field:INFO/NormalMax"
+      },
+      "outcomeTrue": {
+        "nextNode": "exit_lb"
+      },
+      "outcomeMissing": {
+        "nextNode": "exit_vus"
+      },
+      "outcomeFalse": {
+        "nextNode": "exit_vus"
+      }
+    },
     "exit_u1": {
       "description": "Usable: probably",
       "type": "LEAF",
@@ -100,6 +154,21 @@
       "description": "Usable: probably not",
       "type": "LEAF",
       "class": "U3"
+    },
+    "exit_lb": {
+      "description": "STR Likely Benign",
+      "type": "LEAF",
+      "class": "LB"
+    },
+    "exit_vus": {
+      "description": "STR VUS",
+      "type": "LEAF",
+      "class": "VUS"
+    },
+    "exit_lp": {
+      "description": "STR Likely Pathogenic",
+      "type": "LEAF",
+      "class": "LP"
     }
   }
 }

resources/field_metadata.json

@@ -16,6 +16,22 @@
         "YL",
         "MT"
       ]
+    },
+    "ACR": {
+      "label": "Allelic copy ranges",
+      "description": "Straglr: Allelic copy ranges",
+      "numberType": "PER_GENOTYPE",
+      "numberCount": 2,
+      "separator": "-",
+      "type": "FLOAT"
+    },
+    "AC": {
+      "label": "Allelic copies",
+      "description": "Straglr: Allelic copies",
+      "numberType": "PER_GENOTYPE",
+      "numberCount": 2,
+      "separator": "-",
+      "type": "FLOAT"
     }
   },
   "info": {

utils/apptainer/build.sh

@@ -95,8 +95,8 @@ main() {
   images+=("seqtk-1.4")
   images+=("spectre-0.2.1-patched")
   images+=("stranger-0.8.1")
-  images+=("straglr-1.4.4_vip_v3")
-  images+=("vcf-decision-tree-4.1.1")
+  images+=("straglr-1.5.1")
+  images+=("vcf-decision-tree-4.2.0-beta")
   images+=("vcf-inheritance-matcher-3.1.0")
   images+=("vcf-report-7.0.0")
 

utils/apptainer/def/straglr-1.5.1.def

@@ -0,0 +1,24 @@
+Bootstrap: localimage
+From: sif/build/ubuntu-22.04.sif
+
+%post
+    apt-get update
+    apt-get install -qq curl bzip2 gcc make zlib1g-dev libbz2-dev liblzma-dev libdeflate-dev git python3-pip python3
+
+    apt install ncbi-blast+  --assume-yes
+
+    mkdir -p /opt/trf/
+    curl -Ls -o /opt/trf/trf "https://github.com/Benson-Genomics-Lab/TRF/releases/download/v4.09.1/trf409.linux64"
+    chmod +x /opt/trf/trf
+
+    pip install git+https://github.com/bcgsc/[email protected]#egg=straglr
+
+    # cleanup
+    pip cache purge
+    apt-get clean
+
+%environment
+   PATH=$PATH:/opt/trf
+
+%help
+    Long-read-based human genomic short tandem repeat detection.