Add squirrel (#1)

* adding squirrel * fix * wip * make naming of outputs consistent with the rest of the pipeline * make naming of outputs consistent with the rest of the pipeline * Small changes * remove masking from first alignment * add version * update schema and dockerfile sha * update squirrel version * add export * update squirrel defaults * Small changes pre merge * fix * no default clade --------- Co-authored-by: Biowilko <[email protected]>
artic-network · Sep 26, 2024 · 960e885 · 960e885
1 parent d7e41f3
commit 960e885
Show file tree

Hide file tree

Showing 5 changed files with 116 additions and 3 deletions.
diff --git a/modules/squirrel.nf b/modules/squirrel.nf
@@ -0,0 +1,33 @@
+process squirrelAlignmentAndQC {
+
+    label "process_low"
+
+    container "docker.io/articnetworkorg/squirrel@sha256:78921da0c726dd525f5d72db331526975adb634478ee31047c46765bd3d5a64a"
+
+    publishDir "${params.outdir}/${task.process.replaceAll(":","_")}", pattern: "squirrel/**", mode: 'copy', saveAs: {  fn -> fn.replace("squirrel/", "")}
+
+    input:
+        path fasta
+        path refs
+    output:
+        path "squirrel/all_consensus.aln.fasta", emit: alignment
+        path "squirrel/**", emit: all
+        path "squirrel.version", emit: version
+
+  script:
+    extra = ""
+    if ( params.squirrel_assembly_refs )
+        extra += " --assembly-refs ${refs}"
+    if ( params.clade )
+        extra += " --clade ${params.clade}"
+    if ( params.run_phylo )
+        extra += " --run-phylo"
+    if ( params.outgroups )
+        extra += " --outgroups ${params.outgroups}"
+
+    """
+    export XDG_CACHE_HOME=\$PWD/.cache
+    squirrel --version 2>&1 | sed 's/: /,/' > squirrel.version
+    squirrel ${fasta} --no-mask --seq-qc -o squirrel --outfile all_consensus.aln.fasta --tempdir squirrel_tmp -t ${task.cpus} ${extra}
+    """
+}
diff --git a/nextflow.config b/nextflow.config
@@ -5,9 +5,15 @@ manifest {
   description = 'Epi2me compatible Nextflow pipeline for processing ARTIC tiling amplicon Illumina sequencing reads from monkeypox virus (MPXV) samples.'
   mainScript = 'main.nf'
   nextflowVersion = '>=20.01.0'
-  version = '1.1.0'
+  version = '1.2.0'
 }
 
+epi2melabs {
+    tags = 'MPox,artic,amplicon,viruses,public health,illumina'
+    icon = 'faVirusCovid'
+}
+
+
 // Load base.config by default for all pipelines
 includeConfig 'conf/base.conf'
 
@@ -89,7 +95,15 @@ params {
   monochrome_logs = false
   validate_params = true
   show_hidden_params = false
-  schema_ignore_params = 'show_hidden_params,validate_params,monochrome_logs,aws_queue,aws_image_prefix,wf,fastqSearchPath,fastq_exts,illuminaSuffixes,out_dir,illumina-suffixes'
+  schema_ignore_params = 'show_hidden_params,validate_params,monochrome_logs,aws_queue,aws_image_prefix,wf,fastqSearchPath,fastq_exts,illuminaSuffixes,out_dir,illumina-suffixes,run_phylo,no_mask'
+
+  // Squirrel options
+  skip_squirrel = false
+  no_mask = false
+  clade = null
+  run_phylo = false
+  outgroups = null
+  squirrel_assembly_refs = null
 
 }
 

diff --git a/nextflow_schema.json b/nextflow_schema.json
@@ -90,6 +90,52 @@
       },
       "required": ["outdir", "prefix", "store_dir"]
     },
+    "squirrel_options": {
+      "title": "Squirrel Options",
+      "type": "object",
+      "description": "Options related to running Squirrel",
+      "properties": {
+        "skip_squirrel": {
+          "type": "boolean",
+          "default": false,
+          "title": "Skip Squirrel",
+          "description": "Do not run alignment and QC pipeline",
+          "help_text": ""
+        },
+        "clade": {
+          "type": "string",
+          "title": "Clade",
+          "description": "Specify whether the sequencing run is primarily `cladeI` or `cladeII`.",
+          "enum": [
+            "cladei",
+            "cladeia",
+            "cladeib",
+            "cladeii",
+            "cladeiia",
+            "cladeiib"
+          ],
+          "help_text": ""
+        },
+        "outgroups": {
+          "type": "string",
+          "title": "Outgroups",
+          "hidden": true,
+          "description": "Specify which MPXV outgroup(s) in the alignment to use in the phylogeny.",
+          "help_text": "These are required by the phylogenetics module and will get pruned out from the final tree."
+        },
+        "squirrel_assembly_refs": {
+          "type": "string",
+          "format": "path",
+          "title": "Assembly References",
+          "hidden": true,
+          "description": "References to check for `calls to reference` against, used only by the sequence QC process.",
+          "help_text": ""
+        }
+      },
+      "required": ["clade"],
+      "fa_icon": "fas fa-address-card",
+      "help_text": ""
+    },
     "resource_options": {
       "title": "Resource Options",
       "type": "object",
@@ -229,6 +275,12 @@
     {
       "$ref": "#/definitions/output"
     },
+    {
+      "$ref": "#/definitions/squirrel_options"
+    },
+    {
+      "$ref": "#/definitions/resource_options"
+    },
     {
       "$ref": "#/definitions/advanced_options"
     },

diff --git a/test_data/empty.fa b/test_data/empty.fa
diff --git a/workflows/illuminaMpxv.nf b/workflows/illuminaMpxv.nf
@@ -18,6 +18,8 @@ include { makeQCCSV }         from '../modules/qc.nf'
 
 include { collateSamples }    from '../modules/upload.nf'
 
+include { squirrelAlignmentAndQC } from '../modules/squirrel.nf'
+
 workflow prepareReferenceFiles {
 
   c_green = params.monochrome_logs ? '' : "\033[0;32m";
@@ -146,8 +148,19 @@ workflow sequenceAnalysis {
 
       publishQCCSV(qc)
 
+      collateSamples(callConsensusFreebayes.out.consensus.join(ch_filtered_reads))
+
+      callConsensusFreebayes.out.consensus.map{ sampleName,sampleFasta -> sampleFasta }.collectFile(name: "all_consensus.fa").set{ consensus }
+      if ( params.squirrel_assembly_refs ) {
+            refs_ch = channel.fromPath("${params.squirrel_assembly_refs}", checkIfExists:true)
+      } else {
+            refs_ch = channel.fromPath("${projectDir}/test_data/empty.fa", checkIfExists:true)
+      }
+      squirrelAlignmentAndQC(consensus, refs_ch)
+
     emit:
-      qc_pass = callConsensusFreebayes.out.consensus.join(ch_filtered_reads)
+      qc_pass = collateSamples.out
+      alignment = squirrelAlignmentAndQC.out.alignment
 }
 
 workflow mpxvIllumina {
@@ -156,6 +169,7 @@ workflow mpxvIllumina {
 
     main:
       prepareReferenceFiles()
+
       sequenceAnalysis(ch_filePairs, prepareReferenceFiles.out.reference, prepareReferenceFiles.out.bwaIndex, prepareReferenceFiles.out.primers)
 }