Add tb grading rules bed

CHANGELOG.md

@@ -17,7 +17,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Added `get_taxon` to methods
 - Added `streptococcus` and `streptococcus_pyogenes` workflows and profiles to configs
 - Added ska filepath to yaml
-- Added optional read downsampling using seqtk.
+- Added optional read downsampling using seqtk
+- Added `tbGradingRulesBed` to IGV track
 
 ### Fixed
 

Makefile

@@ -752,7 +752,7 @@ $(STREP_CGMLST_DIR)/alleles_rereffed: | $(STREP_CGMLST_DIR)/alleles/index.html
 
 MTUBE_GENOMES_DIR := $(ASSETS_DIR)/genomes/mycobacterium_tuberculosis
 MTUBE_TBDB_DIR := $(ASSETS_DIR)/tbdb
-MTUBE_TBPROFILER_DBS_DIR := $(ASSETS_DIR)/tbprofiler_dbs
+MTUBE_TB_INFO_DIR := $(ASSETS_DIR)/tb_info
 MTUBE_REFSEQ_ACC := GCF_000195955.2
 
 mtuberculosis_all: mtuberculosis_download_reference mtuberculosis_faidx_reference mtuberculosis_bwaidx_reference mtuberculosis_converged_who_fohm_tbdb mtuberculosis_bgzip_bed mtuberculosis_index_bed
@@ -786,10 +786,10 @@ $(MTUBE_GENOMES_DIR)/$(MTUBE_REFSEQ_ACC).fasta.bwt: $(MTUBE_GENOMES_DIR)/$(MTUBE
 
 mtuberculosis_converged_who_fohm_tbdb: $(MTUBE_TBDB_DIR)/converged_who_fohm_tbdb.variables.json
 
-$(MTUBE_TBDB_DIR)/converged_who_fohm_tbdb.variables.json: $(MTUBE_TBPROFILER_DBS_DIR)/converged_who_fohm_tbdb.csv
+$(MTUBE_TBDB_DIR)/converged_who_fohm_tbdb.variables.json: $(MTUBE_TB_INFO_DIR)/csv/converged_who_fohm_tbdb.csv
 	$(call log_message,"Creating WHO FoHM TBDB ...")
 	cd $(MTUBE_TBDB_DIR) \
-	&& cp $(MTUBE_TBPROFILER_DBS_DIR)/converged_who_fohm_tbdb.csv $(MTUBE_TBDB_DIR) \
+	&& cp $(MTUBE_TB_INFO_DIR)/csv/converged_who_fohm_tbdb.csv $(MTUBE_TBDB_DIR) \
 	&& singularity exec --bind $(MNT_ROOT) $(CONTAINER_DIR)/tb-profiler.sif \
 		tb-profiler create_db --prefix converged_who_fohm_tbdb --dir $(MTUBE_TBDB_DIR) \
 		--match_ref $(MTUBE_GENOMES_DIR)/GCF_000195955.2.fasta --csv converged_who_fohm_tbdb.csv \

assets/tb_info/bed/tb_grading_rules.bed

@@ -0,0 +1,5 @@
+NC_000962.3	761325	761405	RRDR	rpob	rifampicin
+NC_000962.3	7563	7565	MFX_gyrA88	gyrA	MFX
+NC_000962.3	7581	7583	MFX_gyrA94	gyrA	MFX
+NC_000962.3	4326003	4327472	LOF_ethA	ethA	ethionamide
+NC_000962.3	4407527	4408201	LOF_gid	gid	streptomycin

configs/nextflow.base.config

@@ -81,11 +81,13 @@ profiles {
 		params.species = 'mycobacterium tuberculosis'
 		params.speciesDir = 'mtuberculosis'
 		params.resistantLociName = 'resistance_loci'
+		params.gradingLociName = 'grading_loci'
 		params.referenceGenome = "${params.root}/assets/genomes/mycobacterium_tuberculosis/GCF_000195955.2.fasta"
 		params.referenceGenomeIdx = "${params.root}/assets/genomes/mycobacterium_tuberculosis/GCF_000195955.2.fasta.fai"
 		params.referenceGenomeGff = "${params.root}/assets/genomes/mycobacterium_tuberculosis/GCF_000195955.2.gff"
 		params.coreLociBed = "${params.root}/assets/cgmlst/mycobacterium_tuberculosis/bed/GCF_000195955.2.bed"
-		params.tbdbBed = "${params.root}/assets/tbprofiler_dbs/bed/converged_who_fohm_tbdb.bed.gz"
+		params.tbGradingRulesBed = "${params.root}/assets/tb_info/bed/tb_grading_rules.bed"
+		params.tbdbBed = "${params.root}/assets/tbdb/converged_who_fohm_tbdb.bed.gz"
 		params.tbdbBedIdx = "${params.tbdbBed}.tbi"
 		params.targetSampleSize = null
 	}

configs/nextflow.ci.config

@@ -85,11 +85,13 @@ profiles {
 		params.species = 'mycobacterium tuberculosis'
 		params.speciesDir = 'mtuberculosis'
 		params.resistantLociName = 'resistance_loci'
+		params.gradingLociName = 'grading_loci'
 		params.referenceGenome = "${params.root}/assets/genomes/mycobacterium_tuberculosis/GCF_000195955.2.fasta"
 		params.referenceGenomeIdx = "${params.root}/assets/genomes/mycobacterium_tuberculosis/GCF_000195955.2.fasta.fai"
 		params.referenceGenomeGff = "${params.root}/assets/genomes/mycobacterium_tuberculosis/GCF_000195955.2.gff"
 		params.coreLociBed = "${params.root}/assets/cgmlst/mycobacterium_tuberculosis/bed/GCF_000195955.2.bed"
-		params.tbdbBed = "${params.root}/assets/tbprofiler_dbs/bed/converged_who_fohm_tbdb.bed.gz"
+		params.tbGradingRulesBed = "${params.root}/assets/tb_info/bed/tb_grading_rules.bed"
+		params.tbdbBed = "${params.root}/assets/tbdb/converged_who_fohm_tbdb.bed.gz"
 		params.tbdbBedIdx = "${params.tbdbBed}.tbi"
 		params.targetSampleSize = null
 	}

configs/nextflow.dev.config

@@ -96,11 +96,13 @@ profiles {
 		params.speciesDir = 'mtuberculosis'
 		params.symlinkDir = "/access/jasen/mtuberculosis/"
 		params.resistantLociName = 'resistance_loci'
+		params.gradingLociName = 'grading_loci'
 		params.referenceGenome = "${params.root}/assets/genomes/mycobacterium_tuberculosis/GCF_000195955.2.fasta"
 		params.referenceGenomeIdx = "${params.root}/assets/genomes/mycobacterium_tuberculosis/GCF_000195955.2.fasta.fai"
 		params.referenceGenomeGff = "${params.root}/assets/genomes/mycobacterium_tuberculosis/GCF_000195955.2.gff"
 		params.coreLociBed = "${params.root}/assets/cgmlst/mycobacterium_tuberculosis/bed/GCF_000195955.2.bed"
-		params.tbdbBed = "${params.root}/assets/tbprofiler_dbs/bed/converged_who_fohm_tbdb.bed.gz"
+		params.tbGradingRulesBed = "${params.root}/assets/tb_info/bed/tb_grading_rules.bed"
+		params.tbdbBed = "${params.root}/assets/tbdb/converged_who_fohm_tbdb.bed.gz"
 		params.tbdbBedIdx = "${params.tbdbBed}.tbi"
 		params.targetSampleSize = null
 	}

configs/nextflow.hopper.config

@@ -95,12 +95,14 @@ profiles {
 		params.species = 'mycobacterium tuberculosis'
 		params.speciesDir = 'mtuberculosis'
 		params.resistantLociName = 'resistance_loci'
+		params.gradingLociName = 'grading_loci'
 		params.symlinkDir = "/access/jasen/mtuberculosis/"
 		params.referenceGenome = "${params.root}/assets/genomes/mycobacterium_tuberculosis/GCF_000195955.2.fasta"
 		params.referenceGenomeIdx = "${params.root}/assets/genomes/mycobacterium_tuberculosis/GCF_000195955.2.fasta.fai"
 		params.referenceGenomeGff = "${params.root}/assets/genomes/mycobacterium_tuberculosis/GCF_000195955.2.gff"
 		params.coreLociBed = "${params.root}/assets/cgmlst/mycobacterium_tuberculosis/bed/GCF_000195955.2.bed"
-		params.tbdbBed = "${params.root}/assets/tbprofiler_dbs/bed/converged_who_fohm_tbdb.bed.gz"
+		params.tbGradingRulesBed = "${params.root}/assets/tb_info/bed/tb_grading_rules.bed"
+		params.tbdbBed = "${params.root}/assets/tbdb/converged_who_fohm_tbdb.bed.gz"
 		params.tbdbBedIdx = "${params.tbdbBed}.tbi"
 		params.targetSampleSize = null
 	}

configs/nextflow.ngp.config

@@ -107,11 +107,13 @@ profiles {
 		params.species = 'mycobacterium tuberculosis'
 		params.speciesDir = 'mtuberculosis'
 		params.resistantLociName = 'resistance_loci'
+		params.gradingLociName = 'grading_loci'
 		params.referenceGenome = "${params.root}/assets/genomes/mycobacterium_tuberculosis/GCF_000195955.2.fasta"
 		params.referenceGenomeIdx = "${params.root}/assets/genomes/mycobacterium_tuberculosis/GCF_000195955.2.fasta.fai"
 		params.referenceGenomeGff = "${params.root}/assets/genomes/mycobacterium_tuberculosis/GCF_000195955.2.gff"
 		params.coreLociBed = "${params.root}/assets/cgmlst/mycobacterium_tuberculosis/bed/GCF_000195955.2.bed"
-		params.tbdbBed = "${params.root}/assets/tbprofiler_dbs/bed/converged_who_fohm_tbdb.bed.gz"
+		params.tbGradingRulesBed = "${params.root}/assets/tb_info/bed/tb_grading_rules.bed"
+		params.tbdbBed = "${params.root}/assets/tbdb/converged_who_fohm_tbdb.bed.gz"
 		params.tbdbBedIdx = "${params.tbdbBed}.tbi"
 		params.targetSampleSize = null
 	}

workflows/bacterial_base.nf

@@ -40,19 +40,19 @@ workflow CALL_BACTERIAL_BASE {
         // Create channel for reads
         ch_raw_input
             .map{ row -> get_reads(row) }
-            .set{ ch_reads }
+            .set{ ch_raw_reads }
 
         // downsample reads
-        seqtk_sample( ch_reads.map(row -> [row[0], row[1], params.targetSampleSize]) ).reads
-            .concat( ch_reads )        // add raw reads channel
+        seqtk_sample( ch_raw_reads.map(row -> [row[0], row[1], params.targetSampleSize]) ).reads
+            .concat( ch_raw_reads )    // add raw reads channel
             .first()                   // if seqtk was not run the first row is the raw reads
-            .set { ch_reads }          // overwrite reads channel
+            .set{ ch_sampled_reads }  // create sampled reads channel
 
         // reads trim and clean and recreate reads channel if the reads were trimmed
-        assembly_trim_clean(ch_reads.join(ch_meta)).set { ch_clean_reads_w_meta }
+        assembly_trim_clean(ch_sampled_reads.join(ch_meta)).set { ch_clean_reads_w_meta }
         Channel.empty()
-            .mix( ch_reads, ch_clean_reads_w_meta )                   // if samples are trimmed
-            .tap{ ch_reads }                                          // overwrite reads channel
+            .mix( ch_sampled_reads, ch_clean_reads_w_meta )           // if samples are trimmed
+            .tap{ ch_reads }                                          // create reads channel
             .join( ch_meta )                                          // add meta info
             .set{ ch_reads_w_meta }                                   // write as temp channel
 

workflows/mycobacterium_tuberculosis.nf

@@ -4,33 +4,24 @@ nextflow.enable.dsl=2
 
 include { get_meta                                  } from '../methods/get_sample_data.nf'
 include { get_reads                                 } from '../methods/get_sample_data.nf'
-include { annotate_delly                        } from '../nextflow-modules/modules/prp/main.nf'
-include { bracken                               } from '../nextflow-modules/modules/bracken/main.nf'
-include { create_analysis_result                } from '../nextflow-modules/modules/prp/main.nf'
-include { add_igv_track as add_variant_igv_track} from '../nextflow-modules/modules/prp/main.nf'
-include { add_igv_track as add_locus_igv_track  } from '../nextflow-modules/modules/prp/main.nf'
-include { create_cdm_input                      } from '../nextflow-modules/modules/prp/main.nf'
-include { create_yaml                           } from '../nextflow-modules/modules/yaml/main.nf'
-include { export_to_cdm                         } from '../nextflow-modules/modules/cmd/main.nf'
-include { kraken                                } from '../nextflow-modules/modules/kraken/main.nf'
-include { mykrobe                               } from '../nextflow-modules/modules/mykrobe/main.nf'
-include { post_align_qc                         } from '../nextflow-modules/modules/prp/main.nf'
-include { snippy                                } from '../nextflow-modules/modules/snippy/main.nf'
-include { tbprofiler as tbprofiler_mergedb      } from '../nextflow-modules/modules/tbprofiler/main.nf'
-include { CALL_BACTERIAL_BASE                   } from '../workflows/bacterial_base.nf'
+include { annotate_delly                            } from '../nextflow-modules/modules/prp/main.nf'
+include { bracken                                   } from '../nextflow-modules/modules/bracken/main.nf'
+include { create_analysis_result                    } from '../nextflow-modules/modules/prp/main.nf'
+include { add_igv_track as add_tbdb_bed_track       } from '../nextflow-modules/modules/prp/main.nf'
+include { add_igv_track as add_grading_bed_track    } from '../nextflow-modules/modules/prp/main.nf'
+include { create_cdm_input                          } from '../nextflow-modules/modules/prp/main.nf'
+include { create_yaml                               } from '../nextflow-modules/modules/yaml/main.nf'
+include { export_to_cdm                             } from '../nextflow-modules/modules/cmd/main.nf'
+include { kraken                                    } from '../nextflow-modules/modules/kraken/main.nf'
+include { mykrobe                                   } from '../nextflow-modules/modules/mykrobe/main.nf'
+include { post_align_qc                             } from '../nextflow-modules/modules/prp/main.nf'
+include { snippy                                    } from '../nextflow-modules/modules/snippy/main.nf'
+include { tbprofiler as tbprofiler_mergedb          } from '../nextflow-modules/modules/tbprofiler/main.nf'
+include { CALL_BACTERIAL_BASE                       } from '../workflows/bacterial_base.nf'
 
 workflow CALL_MYCOBACTERIUM_TUBERCULOSIS {
-    // Create channel for sample metadata
-    Channel.fromPath(params.csv)
-        .splitCsv(header:true)
-        .tap{ ch_raw_input }
-        .map{ row -> get_meta(row) }
-        .set{ ch_meta }
-
-    // Create channel for reads
-    ch_raw_input
-        .map{ row -> get_reads(row) }
-        .set{ ch_reads }
+    // set input data
+    inputSamples = file(params.csv, checkIfExists: true)
 
     // load references 
     referenceGenome = file(params.referenceGenome, checkIfExists: true)
@@ -41,11 +32,12 @@ workflow CALL_MYCOBACTERIUM_TUBERCULOSIS {
     coreLociBed = file(params.coreLociBed, checkIfExists: true)
     tbdbBed = file(params.tbdbBed, checkIfExists: true)
     tbdbBedIdx = file(params.tbdbBedIdx, checkIfExists: true)
+    tbGradingRulesBed = file(params.tbGradingRulesBed, checkIfExists: true)
 
     main:
         ch_versions = Channel.empty()
 
-        CALL_BACTERIAL_BASE( coreLociBed, referenceGenome, referenceGenomeDir, ch_meta, ch_reads )
+        CALL_BACTERIAL_BASE( coreLociBed, referenceGenome, referenceGenomeDir, inputSamples )
 
         CALL_BACTERIAL_BASE.out.assembly.set{ch_assembly}
         CALL_BACTERIAL_BASE.out.reads.set{ch_reads}
@@ -104,10 +96,11 @@ workflow CALL_MYCOBACTERIUM_TUBERCULOSIS {
         }
 
         // Add IGV annotation tracks
-        add_locus_igv_track(create_analysis_result.out.json, params.tbdbBed, params.resistantLociName)
+        add_tbdb_bed_track(create_analysis_result.out.json, params.tbdbBed, params.resistantLociName)
+        add_grading_bed_track(add_tbdb_bed_track.out.json, params.tbGradingRulesBed, params.gradingLociName)
 
         // Create yaml for uploading results to Bonsai
-        create_yaml(add_locus_igv_track.out.json.join(ch_sourmash).join(ch_ska), params.speciesDir)
+        create_yaml(add_grading_bed_track.out.json.join(ch_sourmash).join(ch_ska), params.speciesDir)
 
         ch_quast
             .join(ch_qc)
@@ -125,7 +118,7 @@ workflow CALL_MYCOBACTERIUM_TUBERCULOSIS {
         ch_versions = ch_versions.mix(tbprofiler_mergedb.out.versions)
 
     emit: 
-        pipeline_result = add_locus_igv_track.out.json
+        pipeline_result = add_grading_bed_track.out.json
         cdm             = export_to_cdm.out.cdm
         yaml            = create_yaml.out.yaml
         versions        = ch_versions