Implement ROSE

nf-core · Mar 26, 2024 · 57993c1 · 57993c1
1 parent 6490409
commit 57993c1
Show file tree

Hide file tree

Showing 10 changed files with 777 additions and 4 deletions.
diff --git a/bin/rose.py b/bin/rose.py
diff --git a/conf/igenomes.config b/conf/igenomes.config
@@ -24,6 +24,7 @@ params {
             macs_gsize  = "2.7e9"
             blacklist   = "${projectDir}/assets/blacklists/GRCh37-blacklist.bed"
             pwms        = "${projectDir}/assets/PWMs/Jaspar_Hocomoco_Kellis_human_PSEMs.txt"
+            rose_ucsc   = "https://raw.githubusercontent.com/stjude/ROSE/master/annotation/hg19_refseq.ucsc"
         }
         'GRCh38' {
             fasta       = "${params.igenomes_base}/Homo_sapiens/NCBI/GRCh38/Sequence/WholeGenomeFasta/genome.fa"
@@ -37,6 +38,7 @@ params {
             macs_gsize  = "2.7e9"
             blacklist   = "${projectDir}/assets/blacklists/hg38-blacklist.bed"
             pwms        = "${projectDir}/assets/PWMs/Jaspar_Hocomoco_Kellis_human_PSEMs.txt"
+            rose_ucsc   = "https://raw.githubusercontent.com/stjude/ROSE/master/annotation/hg38_refseq.ucsc"
         }
         'CHM13' {
             fasta       = "${params.igenomes_base}/Homo_sapiens/UCSC/CHM13/Sequence/WholeGenomeFasta/genome.fa"
@@ -59,6 +61,7 @@ params {
             macs_gsize  = "1.87e9"
             blacklist   = "${projectDir}/assets/blacklists/GRCm38-blacklist.bed"
             pwms        = "${projectDir}/assets/PWMs/Jaspar_Hocomoco_Kellis_mouse_PSEMs.txt"
+            rose_ucsc   = "https://raw.githubusercontent.com/stjude/ROSE/master/annotation/mm10_refseq.ucsc"
         }
         'TAIR10' {
             fasta       = "${params.igenomes_base}/Arabidopsis_thaliana/Ensembl/TAIR10/Sequence/WholeGenomeFasta/genome.fa"
@@ -294,6 +297,7 @@ params {
             macs_gsize  = "2.7e9"
             blacklist   = "${projectDir}/assets/blacklists/hg38-blacklist.bed"
             pwms        = "${projectDir}/assets/PWMs/Jaspar_Hocomoco_Kellis_human_PSEMs.txt"
+            rose_ucsc   = "https://raw.githubusercontent.com/stjude/ROSE/master/annotation/hg38_refseq.ucsc"
         }
         'hg19' {
             fasta       = "${params.igenomes_base}/Homo_sapiens/UCSC/hg19/Sequence/WholeGenomeFasta/genome.fa"
@@ -308,6 +312,7 @@ params {
             macs_gsize  = "2.7e9"
             blacklist   = "${projectDir}/assets/blacklists/hg19-blacklist.bed"
             pwms        = "${projectDir}/assets/PWMs/Jaspar_Hocomoco_Kellis_human_PSEMs.txt"
+            rose_ucsc   = "https://raw.githubusercontent.com/stjude/ROSE/master/annotation/hg19_refseq.ucsc"
         }
         'mm10' {
             fasta       = "${params.igenomes_base}/Mus_musculus/UCSC/mm10/Sequence/WholeGenomeFasta/genome.fa"
@@ -322,6 +327,7 @@ params {
             macs_gsize  = "1.87e9"
             blacklist   = "${projectDir}/assets/blacklists/mm10-blacklist.bed"
             pwms        = "${projectDir}/assets/PWMs/Jaspar_Hocomoco_Kellis_mouse_PSEMs.txt"
+            rose_ucsc   = "https://raw.githubusercontent.com/stjude/ROSE/master/annotation/mm10_refseq.ucsc"
         }
         'bosTau8' {
             fasta       = "${params.igenomes_base}/Bos_taurus/UCSC/bosTau8/Sequence/WholeGenomeFasta/genome.fa"

diff --git a/conf/modules.config b/conf/modules.config
@@ -61,6 +61,25 @@ process {
     withName: ".*:CHROMHMM:REHEADER_CONTROL" {
         ext.prefix = {"${meta.id}_control"}
     }
+
+    withName: BED_TO_GFF {
+        ext.args = {"'BEGIN{FS=\"\\t\";OFS=\"\\t\"} { print \$1, \"bed2gff\", \"region\", \$2+1, \$3, \".\", \".\", \".\", \".\"}'"}
+        ext.prefix = {"$meta.id"}
+        ext.suffix = "gff"
+    }
+
+    withName: REFORMAT_GFF {
+        ext.args = {"'BEGIN{FS=\"\\t\";OFS=\"\\t\"} {if(!match(\$1, /^chr/)) \$1=\"chr\"\$1; \$2=\"seq_\"NR; print \$1, \$2, \"\", \$4, \$5, \"\", \$7, \"\", \$2}'"}
+        ext.prefix = {"${meta.id}_reformatted"}
+        ext.suffix = "gff"
+    }
+
+    withName: ROSE_OUTPUT_TO_BED {
+        ext.args = {"'BEGIN{FS=\"\\t\";OFS=\"\\t\"} {print \$1, \$4-1, \$5}'"}
+        ext.prefix = {"$meta.id"}
+        ext.suffix = "bed"
+    }
+
     withName: ".*DYNAMITE:FILTER" {
         ext.args = {"'BEGIN{OFS=\"\\t\"} NR==1 || (\$2 >= ${params.dynamite_min_regression} || \$2 <= -${params.dynamite_min_regression} )'"}
         ext.prefix = {"${meta.id}.filtered"}

diff --git a/main.nf b/main.nf
@@ -34,6 +34,7 @@ params.fasta     = getGenomeAttribute('fasta')
 params.gtf       = getGenomeAttribute('gtf')
 params.blacklist = getGenomeAttribute('blacklist')
 params.pwms      = getGenomeAttribute('pwms')
+params.rose_ucsc = getGenomeAttribute('rose_ucsc')
 
 /*
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -87,6 +88,7 @@ workflow NFCORE_TFACTIVITY {
         samplesheet_bam,
         PREPARE_GENOME.out.chrom_sizes,
         params.chromhmm_states,
+        params.rose_ucsc,
         params.window_size,
         params.decay,
         params.merge_samples,

diff --git a/modules/local/rose/main.nf b/modules/local/rose/main.nf
@@ -0,0 +1,31 @@
+process ROSE {
+    tag "$meta.id"
+    label 'process_single'
+
+    conda "conda-forge::mulled-v2-2076f4a3fb468a04063c9e6b7747a630abb457f6==fccb0c41a243c639e11dd1be7b74f563e624fcca-0"
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/mulled-v2-2076f4a3fb468a04063c9e6b7747a630abb457f6:fccb0c41a243c639e11dd1be7b74f563e624fcca-0':
+        'biocontainers/mulled-v2-2076f4a3fb468a04063c9e6b7747a630abb457f6:fccb0c41a243c639e11dd1be7b74f563e624fcca-0' }"
+
+    input:
+    tuple val(meta), path(gff)
+    path ucsc_file
+
+    output:
+    tuple val(meta), path("${gff.baseName}_STITCHED.gff")
+
+    script:
+    """
+    rose.py \
+    -g ${ucsc_file} \
+    -i ${gff} \
+    -o ${gff.baseName}_STITCHED.gff \
+    -s 12500 \
+    -t 2500
+    """
+
+    stub:
+    """
+    touch "${gff.baseName}_STITCHED.gff"
+    """
+}
diff --git a/nextflow_schema.json b/nextflow_schema.json
@@ -253,6 +253,16 @@
                     "help_text": "This parameter is *mandatory* if `--genome` is not specified.",
                     "fa_icon": "far fa-file-code"
                 },
+                "rose_ucsc": {
+                    "type": "string",
+                    "format": "file-path",
+                    "exists": true,
+                    "mimetype": "text/plain",
+                    "pattern": "^\\S+\\.ucsc$",
+                    "description": "Path to ROSE UCSC file.",
+                    "help_text": "This parameter is *mandatory* if `--genome` is not specified and input_bam is defined.",
+                    "fa_icon": "far fa-file-code"
+                },
                 "igenomes_ignore": {
                     "type": "boolean",
                     "description": "Do not load the iGenomes reference config.",

diff --git a/subworkflows/local/chromhmm.nf b/subworkflows/local/chromhmm.nf
@@ -34,6 +34,9 @@ workflow CHROMHMM {
     ch_joined  = ch_signal.join(ch_control)
     ch_mixed   = ch_signal.mix(ch_control)
 
+    ch_versions = ch_versions.mix(REHEADER_SIGNAL.out.versions)
+    ch_versions = ch_versions.mix(REHEADER_CONTROL.out.versions)
+
     ch_table   = ch_joined .map{meta, signal, control -> [meta.condition, meta.antibody, signal.name, control.name]}
                                     .collectFile() {
                                         ["cellmarkfiletable.tsv", it.join("\t") + "\n"]
@@ -50,8 +53,6 @@ workflow CHROMHMM {
         n_states
     )
 
-    LEARN_MODEL.out.transpose().view()
-
     GET_RESULTS(LEARN_MODEL.out.transpose()
                                 .map{meta, emmisions, bed ->
                                     [meta + [id: bed.simpleName.split("_")[0]],
@@ -62,6 +63,6 @@ workflow CHROMHMM {
     emit:
     enhancers = ch_enhancers
 
-    versions = ch_versions                     // channel: [ versions.yml ]
+    versions = ch_versions
 }
 
diff --git a/subworkflows/local/peaks.nf b/subworkflows/local/peaks.nf
@@ -12,6 +12,7 @@ include { COMBINE_TABLES as AFFINITY_SUM  } from '../../modules/local/combine_ta
 include { FOOTPRINTING               } from './footprinting'
 include { MERGE_SAMPLES              } from './merge_samples'
 include { CHROMHMM                   } from './chromhmm'
+include { ROSE                       } from './rose'
 
 workflow PEAKS {
 
@@ -31,6 +32,7 @@ workflow PEAKS {
     ch_samplesheet_bam
     chrom_sizes
     chromhmm_states
+    rose_ucsc
 
     main:
 
@@ -60,6 +62,10 @@ workflow PEAKS {
     }
 
     CHROMHMM(ch_samplesheet_bam, chrom_sizes, chromhmm_states)
+    ROSE(CHROMHMM.out.enhancers, rose_ucsc)
+
+    ch_versions = ch_versions.mix(CHROMHMM.out.versions)
+    ch_versions = ch_versions.mix(ROSE.out.versions)
 
     FILTER_PWMS(tfs, pwms)
 

diff --git a/subworkflows/local/rose.nf b/subworkflows/local/rose.nf
@@ -0,0 +1,30 @@
+include { GAWK as BED_TO_GFF         } from "../../modules/nf-core/gawk"
+include { GAWK as REFORMAT_GFF       } from "../../modules/nf-core/gawk"
+include { ROSE as RUN_ROSE           } from "../../modules/local/rose"
+include { GAWK as ROSE_OUTPUT_TO_BED } from "../../modules/nf-core/gawk"
+
+workflow ROSE {
+    take:
+    ch_bed
+    ucsc_file
+
+    main:
+
+    ch_versions = Channel.empty()
+
+    BED_TO_GFF(ch_bed, [])
+    REFORMAT_GFF(BED_TO_GFF.out.output, [])
+
+    RUN_ROSE(REFORMAT_GFF.out.output, ucsc_file)
+    ROSE_OUTPUT_TO_BED(RUN_ROSE.out, [])
+
+    ch_versions = ch_versions.mix(BED_TO_GFF.out.versions)
+    ch_versions = ch_versions.mix(REFORMAT_GFF.out.versions)
+    ch_versions = ch_versions.mix(ROSE_OUTPUT_TO_BED.out.versions)
+
+    emit:
+    enhancers = ROSE_OUTPUT_TO_BED.out.output
+
+    versions = ch_versions
+}
+
diff --git a/workflows/tfactivity.nf b/workflows/tfactivity.nf
@@ -38,6 +38,7 @@ workflow TFACTIVITY {
     ch_samplesheet_bam
     chrom_sizes
     chromhmm_states
+    rose_ucsc
 
     window_size
     decay
@@ -98,7 +99,8 @@ workflow TFACTIVITY {
         affinity_agg_method,
         ch_samplesheet_bam,
         chrom_sizes,
-        chromhmm_states
+        chromhmm_states,
+        rose_ucsc
     )
 
     DYNAMITE(