Calc frags now working (ad9d5c6c) · Commits · Chaos / Cutandrun

conf/modules.config

+86 −67

Original line number	Diff line number	Diff line
		@@ -468,7 +468,7 @@ if(params.run_peak_calling) {

		if(params.run_peak_calling && 'seacr' in params.callers) {
		process {
		withName: 'NFCORE_CUTANDRUN:CUTANDRUN:SEACR_.*' {
		withName: '.:CUTANDRUN:SEACR_.' {
		ext.args = "non stringent"
		ext.prefix = { "${meta.id}.seacr.peaks.bed" }
		publishDir = [
		@@ -483,7 +483,7 @@ if(params.run_peak_calling && 'seacr' in params.callers) {

		if(params.run_peak_calling && 'macs2' in params.callers) {
		process {
		withName: 'NFCORE_CUTANDRUN:CUTANDRUN:MACS2_.*' {
		withName: '.:CUTANDRUN:MACS2_.' {
		ext.args = "-p ${params.macs2_pvalue}"
		ext.prefix = { "${meta.id}.macs2.peaks.bed" }
		publishDir = [
		@@ -504,7 +504,7 @@ if(params.run_peak_calling && 'macs2' in params.callers) {

		if(params.run_peak_calling) {
		process {
		withName: 'NFCORE_CUTANDRUN:CUTANDRUN:AWK_NAME_PEAK_BED' {
		withName: '.*:AWK_NAME_PEAK_BED' {
		ext.command = "'{OFS = \"\\t\"} {print \$0, FILENAME}'"
		ext.ext = "bed"
		publishDir = [
		@@ -544,7 +544,7 @@ if(params.run_peak_calling) {

		if(params.run_peak_calling && !params.run_consensus_all) {
		process {
		withName: 'NFCORE_CUTANDRUN:CUTANDRUN:CONSENSUS_PEAKS:AWK' {
		withName: '.*:CONSENSUS_PEAKS:AWK' {
		ext.command = "' \$10 >= 1 {print \$0}'"
		ext.ext = "bed"
		publishDir = [
		@@ -559,7 +559,7 @@ if(params.run_peak_calling && !params.run_consensus_all) {

		if(params.run_peak_calling && params.run_consensus_all) {
		process {
		withName: 'NFCORE_CUTANDRUN:CUTANDRUN:CONSENSUS_PEAKS_ALL:AWK' {
		withName: '.*:CONSENSUS_PEAKS_ALL:AWK' {
		ext.command = "' \$10 >= " + params.replicate_threshold.toString() + " {print \$0}'"
		ext.ext = "bed"
		publishDir = [
		@@ -572,6 +572,86 @@ if(params.run_peak_calling && params.run_consensus_all) {
		}
		}

		/*
		========================================================================================
		CALCULATE FRAGMENTS
		========================================================================================
		*/

		if(params.run_peak_calling) {
		process {
		withName: '.*:CALCULATE_FRAGMENTS:SAMTOOLS_VIEW' {
		ext.args = "-F 0x04 -b"
		ext.prefix = { "${meta.id}.mapped" }
		publishDir = [
		enabled: false
		]
		}

		withName: '.*:CALCULATE_FRAGMENTS:SAMTOOLS_SORT' {
		ext.args = "-n"
		ext.prefix = { "${meta.id}.mapped.sorted" }
		publishDir = [
		enabled: false
		]
		}

		withName: '.*:CALCULATE_FRAGMENTS:BEDTOOLS_BAMTOBED' {
		ext.args = "-bedpe"
		publishDir = [
		enabled: false
		]
		}

		withName: '.*:CALCULATE_FRAGMENTS:AWK' {
		ext.command = "'\$1==\$4 && \$6-\$2 < 1000 {print \$0}'"
		ext.suffix = ".filt"
		ext.ext = "bed"
		publishDir = [
		enabled: false
		]
		}

		withName: '.*:CALCULATE_FRAGMENTS:CUT' {
		ext.args = "-f 1,2,6"
		ext.command = "\| sort -T '.' -k1,1 -k2,2n -k3,3n"
		ext.suffix = ".frags"
		ext.ext = "bed"
		publishDir = [
		path: { "${params.outdir}/03_peak_calling/06_fragments" },
		mode: 'copy',
		saveAs: { filename -> filename.equals('versions.yml') ? null : filename },
		enabled: true
		]
		}

		withName: 'NFCORE_CUTANDRUN:CUTANDRUN:AWK_FRAG_BIN' {
		ext.args = "-v w=500"
		ext.command = "'{print \$1, int((\$2 + \$3)/(2w))w + w/2, FILENAME}'"
		ext.command2 = "\| sort -T '.' -k1,1V -k2,2n \| uniq -c \| awk -v OFS=\"\\t\" '{print \$2, \$3, \$1, \$4}' \| sort -T '.' -k1,1V -k2,2n"
		ext.suffix = ".frags.bin500"
		publishDir = [
		path: { "${params.outdir}/03_peak_calling/06_fragments" },
		mode: 'copy',
		saveAs: { filename -> filename.equals('versions.yml') ? null : filename },
		enabled: true
		]
		}

		withName: 'NFCORE_CUTANDRUN:CUTANDRUN:SAMTOOLS_CUSTOMVIEW' {
		ext.args = "-F 0x04"
		ext.args2 = "awk -F'\\t' 'function abs(x){return ((x < 0.0) ? -x : x)} {print abs(\$9)}' \| sort -T '.' \| uniq -c \| awk -v OFS=\"\\t\" '{print \$2, \$1/2}'"
		ext.suffix = ".frags.len"
		publishDir = [
		path: { "${params.outdir}/03_peak_calling/06_fragments" },
		mode: 'copy',
		saveAs: { filename -> filename.equals('versions.yml') ? null : filename },
		enabled: true
		]
		}
		}
		}

		// //TODO: PEAK PLOTTING options - infact all reporting options - how t


		@@ -653,64 +733,3 @@ if(params.run_peak_calling && params.run_consensus_all) {
		// args = "-v"
		// publish_dir = "04_reporting/multiqc"
		// }
		No newline at end of file

		// /*
		// ========================================================================================
		// CALCULATE FRAGMENTS
		// ========================================================================================
		// */

		// "calc_frag_samtools" {
		// suffix = ".mapped"
		// publish_files = false
		// }

		// "calc_frag_samtools_view" {
		// args = "-F 0x04 -b"
		// suffix = ".mapped"
		// publish_files = false
		// }

		// "calc_frag_samtools_sort" {
		// args = "-n"
		// suffix = ".sorted"
		// publish_files = false
		// }

		// "calc_frag_bamtobed" {
		// args = "-bedpe"
		// publish_files = false
		// }

		// "calc_frag_awk" {
		// suffix = ".clean"
		// ext = "bed"
		// command = "'\$1==\$4 && \$6-\$2 < 1000 {print \$0}'"
		// publish_files = false
		// }

		// "calc_frag_cut" {
		// args = "-f 1,2,6"
		// suffix = ".frags"
		// ext = "bed"
		// command = "\| sort -T '.' -k1,1 -k2,2n -k3,3n"
		// publish_dir = "03_peak_calling/06_fragments"
		// }

		// "awk_frag_bin" {
		// args = "-v w=500"
		// suffix = ".frags.bin500"
		// ext = "bed"
		// publish_dir = "03_peak_calling/06_fragments"
		// command = "'{print \$1, int((\$2 + \$3)/(2w))w + w/2, FILENAME}'"
		// command2 = "\| sort -T '.' -k1,1V -k2,2n \| uniq -c \| awk -v OFS=\"\\t\" '{print \$2, \$3, \$1, \$4}' \| sort -T '.' -k1,1V -k2,2n"
		// }

		// "samtools_frag_len" {
		// args = "-F 0x04"
		// args2 = "awk -F'\\t' 'function abs(x){return ((x < 0.0) ? -x : x)} {print abs(\$9)}' \| sort -T '.' \| uniq -c \| awk -v OFS=\"\\t\" '{print \$2, \$1/2}'"
		// suffix = ".frag_len"
		// publish_dir = "03_peak_calling/06_fragments"
		// }
		// }
		// }

modules/local/linux/awk.nf

+2 −2

Original line number	Diff line number	Diff line
		@@ -4,8 +4,8 @@ process AWK {

		conda (params.enable_conda ? "conda-forge::gawk=5.1.0" : null)
		container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
		'https://depot.galaxyproject.org/singularity/gawk:5.1.0' :
		'quay.io/biocontainers/gawk:5.1.0' }"
		'https://containers.biocontainers.pro/s3/SingImgsRepo/biocontainers/v1.2.0_cv1/biocontainers_v1.2.0_cv1.img' :
		'biocontainers/biocontainers:v1.2.0_cv1' }"

		input:
		tuple val(meta), path(input)

workflows/cutandrun.nf

+17 −18

Original line number	Diff line number	Diff line
		@@ -100,10 +100,10 @@ if ((caller_list + callers).unique().size() != caller_list.size()) {
		*/
		include { INPUT_CHECK } from "../subworkflows/local/input_check"
		include { AWK as AWK_NAME_PEAK_BED } from "../modules/local/linux/awk"
		include { AWK as AWK_FRAG_BIN } from "../modules/local/linux/awk"
		include { SAMTOOLS_CUSTOMVIEW } from "../modules/local/samtools_custom_view"
		// include { IGV_SESSION } from "../modules/local/python/igv_session" addParams( options: modules["igv"] )
		// include { AWK as AWK_EDIT_PEAK_BED } from "../modules/local/linux/awk" addParams( options: modules["awk_edit_peak_bed"] )
		// include { AWK as AWK_FRAG_BIN } from "../modules/local/linux/awk" addParams( options: modules["awk_frag_bin"] )
		// include { SAMTOOLS_CUSTOMVIEW } from "../modules/local/samtools_custom_view" addParams( options: modules["samtools_frag_len"] )
		// include { CALCULATE_FRIP } from "../modules/local/modules/calculate_frip/main" addParams( options: modules["calc_frip"] )
		// include { CUT as CUT_CALC_REPROD } from "../modules/local/linux/cut" addParams( options: modules["calc_peak_repro_cut"] )
		// include { CALCULATE_PEAK_REPROD } from "../modules/local/modules/calculate_peak_reprod/main" addParams( options: modules["calc_peak_repro"] )
		@@ -123,7 +123,7 @@ include { ANNOTATE_META_AWK as ANNOTATE_BT2_SPIKEIN_META } from "../subworkflows
		include { ANNOTATE_META_AWK as ANNOTATE_DEDUP_META } from "../subworkflows/local/annotate_meta_awk"
		include { CONSENSUS_PEAKS } from "../subworkflows/local/consensus_peaks"
		include { CONSENSUS_PEAKS as CONSENSUS_PEAKS_ALL } from "../subworkflows/local/consensus_peaks"
		// include { CALCULATE_FRAGMENTS } from "../subworkflows/local/calculate_fragments" addParams( samtools_options: modules["calc_frag_samtools"], samtools_view_options: modules["calc_frag_samtools_view"], samtools_sort_options: modules["calc_frag_samtools_sort"], bamtobed_options: modules["calc_frag_bamtobed"], awk_options: modules["calc_frag_awk"], cut_options: modules["calc_frag_cut"] )
		include { CALCULATE_FRAGMENTS } from "../subworkflows/local/calculate_fragments"
		// include { ANNOTATE_META_CSV as ANNOTATE_FRIP_META } from "../subworkflows/local/annotate_meta_csv" addParams( options: modules["meta_csv_frip_options"] )
		// include { ANNOTATE_META_CSV as ANNOTATE_PEAK_REPRO_META } from "../subworkflows/local/annotate_meta_csv" addParams( options: modules["meta_csv_peak_repro_options"] )

		@@ -727,39 +727,38 @@ workflow CUTANDRUN {
		* - Keep the read pairs that are on the same chromosome and fragment length less than 1000bp
		* - Only extract the fragment related columns using cut
		*/
		// CALCULATE_FRAGMENTS (
		// ch_samtools_bam
		// )
		// ch_software_versions = ch_software_versions.mix(CALCULATE_FRAGMENTS.out.versions)
		CALCULATE_FRAGMENTS (
		ch_samtools_bam
		)
		ch_software_versions = ch_software_versions.mix(CALCULATE_FRAGMENTS.out.versions)
		//EXAMPLE CHANNEL STRUCT: NO CHANGE
		//CALCULATE_FRAGMENTS.out.bed \| view

		/*
		* MODULE: Bin the fragments into 500bp bins ready for downstream reporting
		*/
		// AWK_FRAG_BIN(
		// CALCULATE_FRAGMENTS.out.bed
		// )
		AWK_FRAG_BIN(
		CALCULATE_FRAGMENTS.out.bed
		)
		//AWK_FRAG_BIN.out.file \| view


		/*
		* CHANNEL: Combine bam and bai files on id
		*/
		// ch_samtools_bam
		// .map { row -> [row[0].id, row ].flatten()}
		// .join ( ch_samtools_bai.map { row -> [row[0].id, row ].flatten()} )
		// .map { row -> [row[1], row[2], row[4]] }
		// .set { ch_bam_bai }
		ch_samtools_bam.map { row -> [row[0].id, row ].flatten()}
		.join ( ch_samtools_bai.map { row -> [row[0].id, row ].flatten()} )
		.map { row -> [row[1], row[2], row[4]] }
		.set { ch_bam_bai }
		// EXAMPLE CHANNEL STRUCT: [[META], BAM, BAI]
		//ch_bam_bai \| view

		/*
		* MODULE: Calculate fragment lengths
		*/
		// SAMTOOLS_CUSTOMVIEW (
		// ch_bam_bai
		// )
		SAMTOOLS_CUSTOMVIEW (
		ch_bam_bai
		)
		// ch_software_versions = ch_software_versions.mix(SAMTOOLS_CUSTOMVIEW.out.versions)
		//SAMTOOLS_CUSTOMVIEW.out.tsv \| view
		}

Admin message