Updates to prep for coverage subworkflow

sanger-tol · Dec 1, 2022 · c045758 · c045758
1 parent 250680b
commit c045758
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 6 deletions.
diff --git a/assets/treeval_test.yaml b/assets/treeval_test.yaml
@@ -7,6 +7,10 @@ assembly:
   dbVersion: "1"
   gevalType: DTOL
 reference_file: /lustre/scratch123/tol/teams/grit/geval_pipeline/geval_runs/DTOL/nxOscDoli1_1/data/DTOL_nxOscDoli1_1_FULL.fa
+assembly_reads:
+  pacbio: path
+  hic: path
+  supplementary: path|na
 alignment:
   data_dir: /nfs/team135/dp24/treeval_testdata/gene_alignment_data/
   common_name: "" # For future implementation (adding bee, wasp, ant etc)
@@ -18,5 +22,4 @@ self_comp:
 synteny:
   synteny_genome_path: "/nfs/team135/dp24/treeval_testdata/synteny_data"
 outdir: "NEEDS TESTING"
-intron:
-  size: "50k"
+intron: "50k"
diff --git a/subworkflows/local/yaml_input.nf b/subworkflows/local/yaml_input.nf
@@ -16,6 +16,7 @@ workflow INPUT_READ {
         .flatten()
         .multiMap { data -> 
                 assembly:               ( data.assembly )
+                assembly_reads:         ( data.assem_reads )
                 reference:              ( file(data.reference_file) )
                 alignment:              ( data.alignment )
                 self_comp:              ( data.self_comp )
@@ -38,6 +39,15 @@ workflow INPUT_READ {
             }
         .set { assembly_data }
 
+    group
+        .assem_reads
+        .multiMap { data -> 
+            pacbio:                     data.pacbio
+            hic:                        data.hic
+            supplement:                 data.supplementary
+        }
+        .set { assembly_reads }
+
     group
         .alignment
         .multiMap { data ->
@@ -71,6 +81,10 @@ workflow INPUT_READ {
     assembly_dbVer                   = assembly_data.dbVersion
     assembly_gtype                   = assembly_data.gevalType
 
+    pacbio_reads                     = assembly_reads.pacbio
+    hic_reads                        = assembly_reads.hic
+    supp_reads                       = assembly_reads.supplement
+
     reference                        = group.reference
 
     align_data_dir                   = alignment_data.data_dir

diff --git a/workflows/treeval.nf b/workflows/treeval.nf
@@ -28,6 +28,7 @@ include { INSILICO_DIGEST   } from '../subworkflows/local/insilico_digest'
 include { GENE_ALIGNMENT    } from '../subworkflows/local/gene_alignment'
 include { SELFCOMP          } from '../subworkflows/local/selfcomp'
 include { SYNTENY           } from '../subworkflows/local/synteny'
+// include { LONGREAD_COVERAGE } from '../subworkflows/local/longread_coverage'
 
 /*
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -84,7 +85,7 @@ workflow TREEVAL {
 
 
     //
-    //SUBWORKFLOW: 
+    // SUBWORKFLOW: 
     //
     ch_enzyme = Channel.of( "bspq1","bsss1","DLE1" )
     INSILICO_DIGEST ( INPUT_READ.out.assembly_id,
@@ -95,7 +96,7 @@ workflow TREEVAL {
     ch_versions = ch_versions.mix(INSILICO_DIGEST.out.versions)
 
     //
-    //SUBWORKFLOW: Takes input fasta to generate BB files containing alignment data
+    // SUBWORKFLOW: Takes input fasta to generate BB files containing alignment data
     //
     INPUT_READ.out.intron_size.view()
 
@@ -111,7 +112,7 @@ workflow TREEVAL {
     ch_versions = ch_versions.mix(GENERATE_GENOME.out.versions)
 
     //
-    //SUBWORKFLOW: 
+    // SUBWORKFLOW: 
     //
     SELFCOMP ( GENERATE_GENOME.out.reference_tuple,
                GENERATE_GENOME.out.dot_genome,
@@ -121,13 +122,20 @@ workflow TREEVAL {
     ch_versions = ch_versions.mix(SELFCOMP.out.versions)
 
     //
-    //SUBWORKFLOW: 
+    // SUBWORKFLOW: 
     //
     SYNTENY ( GENERATE_GENOME.out.reference_tuple, 
               INPUT_READ.out.synteny_path,  
               INPUT_READ.out.assembly_classT)
     ch_versions = ch_versions.mix(SYNTENY.out.versions)
 
+    //
+    // SUBWORKFLOW: 
+    //
+    // LONGREAD_COVERAGE (  GENERATE_GENOME.out.reference_tuple,
+    //                      PACBIO.READ.DIRECTORY,
+    //                      INPUT_READ.out.sizeClass )
+
     //
     // SUBWORKFLOW: Collates version data from prior subworflows
     //