diff --git a/CHANGELOG.md b/CHANGELOG.md
index fd4557dde46498c38991bb1c9bc8a174e831e772..0f84a8a58653183b7901828c520e1adfb2447894 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,6 +5,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
## v1.3.0dev
+* Add `--hicpro_maps` options to generate the raw and normalized HiC-Pro maps. The default is now to use cooler
* Add HiCExplorer distance decay quality control
* Add HiCExplorer TADs calling
* Add insulation score TADs calling
diff --git a/main.nf b/main.nf
index e45087a3d979c6a75766868af11fd8de0a5c1688..177050a54bf87af6e80d74783c5b344119157021 100644
--- a/main.nf
+++ b/main.nf
@@ -730,7 +730,7 @@ process merge_stats {
set val(prefix), file(fstat) from all_mapstat.groupTuple().concat(all_pairstat.groupTuple(), all_rsstat.groupTuple())
output:
- file("mstats/") into all_mstats
+ file("mstats/${sample}/*") into all_mstats
script:
sample = prefix.toString() - ~/(_R1|_R2|_val_1|_val_2|_1|_2)/
@@ -745,7 +745,7 @@ process merge_stats {
/*
* HiC-Pro build matrix processes
- * ONGOING VALIDATION - TO REPLACED BY COOLER ?
+ * kept for backward compatibility
*/
@@ -755,7 +755,7 @@ process build_contact_maps{
publishDir "${params.outdir}/hicpro/matrix/raw", mode: params.publish_dir_mode
when:
- !params.skip_maps
+ !params.skip_maps && params.hicpro_maps
input:
set val(sample), file(vpairs), val(mres) from ch_vpairs.combine(map_res)
@@ -776,14 +776,14 @@ process run_ice{
publishDir "${params.outdir}/hicpro/matrix/iced", mode: params.publish_dir_mode
when:
- !params.skip_maps && !params.skip_ice
+ !params.skip_maps && !params.skip_balancing && params.hicpro_maps
input:
set val(sample), val(res), file(rmaps), file(bed) from raw_maps
output:
- set val(sample), val(res), file("*iced.matrix"), file(bed) into iced_maps_4h5, iced_maps_4cool
- file ("*.biases") into iced_bias
+ set val(sample), val(res), file("*iced.matrix"), file(bed) into hicpro_iced_maps
+ file ("*.biases") into hicpro_iced_bias
script:
prefix = rmaps.toString() - ~/(\.matrix)?$/
@@ -896,31 +896,6 @@ process cooler_zoomify {
}
-/*
- * Create h5 file
-
-process convert_to_h5 {
- tag "$sample"
- label 'process_medium'
- publishDir "${params.outdir}/contact_maps/norm/h5", mode: 'copy'
-
- input:
- set val(sample), val(res), file(maps) from norm_cool_maps_h5
-
- output:
- set val(sample), val(res), file("*.h5") into h5maps_ddecay, h5maps_ccomp, h5maps_tads
-
- script:
- """
- hicConvertFormat --matrices ${maps} \
- --outFileName ${maps.baseName}.h5 \
- --resolution ${res} \
- --inputFormat cool \
- --outputFormat h5 \
- """
-}
-*/
-
/****************************************************
* DOWNSTREAM ANALYSIS
*/
@@ -946,7 +921,7 @@ process dist_decay {
!params.skip_dist_decay
input:
- set val(sample), val(res), file(h5mat), val(r) from chddecay
+ set val(sample), val(res), file(maps), val(r) from chddecay
output:
file("*_distcount.txt")
@@ -955,9 +930,9 @@ process dist_decay {
script:
"""
- hicPlotDistVsCounts --matrices ${h5mat} \
- --plotFile ${h5mat.baseName}_distcount.png \
- --outFileData ${h5mat.baseName}_distcount.txt
+ hicPlotDistVsCounts --matrices ${maps} \
+ --plotFile ${maps.baseName}_distcount.png \
+ --outFileData ${maps.baseName}_distcount.txt
"""
}
diff --git a/nextflow.config b/nextflow.config
index c5b39add6663b624d616043b62bc5fa75a749b30..ecd633f8f8239746ee219fc869792354b6c2ed51 100644
--- a/nextflow.config
+++ b/nextflow.config
@@ -62,6 +62,7 @@ params {
// Contact maps
bin_size = '1000000'
res_zoomify = '5000'
+ hicpro_maps = false
ice_max_iter = 100
ice_filter_low_count_perc = 0.02
ice_filter_high_count_perc = 0
@@ -75,7 +76,6 @@ params {
// Workflow
skip_maps = false
- skip_ice = false
skip_balancing = false
skip_mcool = false
skip_dist_decay = false
diff --git a/nextflow_schema.json b/nextflow_schema.json
index 92748776bce1616a11b79c049c14601d93f91bc4..93591e75333352a38532366e07058e6c1aaa3595 100644
--- a/nextflow_schema.json
+++ b/nextflow_schema.json
@@ -162,12 +162,12 @@
"bwt2_opts_end2end": {
"type": "string",
"default": "'--very-sensitive -L 30 --score-min L,-0.6,-0.2 --end-to-end --reorder'",
- "description": "Option for end-to-end bowtie mapping"
+ "description": "Option for HiC-Pro end-to-end bowtie mapping"
},
"bwt2_opts_trimmed": {
"type": "string",
"default": "'--very-sensitive -L 20 --score-min L,-0.6,-0.2 --end-to-end --reorder'",
- "description": "Option for trimmed reads mapping"
+ "description": "Option for HiC-Pro trimmed reads mapping"
},
"save_aligned_intermediates": {
"type": "boolean",
@@ -212,7 +212,7 @@
}
}
},
- "contact_maps_options": {
+ "contact_maps": {
"title": "Contact maps",
"type": "object",
"description": "Options to build Hi-C contact maps",
@@ -224,24 +224,28 @@
"default": "'1000000,500000'",
"description": "Resolution to build the maps (comma separated)"
},
+ "hicpro_maps": {
+ "type": "string",
+ "description": "Generate raw and normalized contact maps with HiC-Pro"
+ },
"ice_filter_low_count_perc": {
"type": "number",
"default": 0.02,
- "description": "Filter low counts rows before normalization"
+ "description": "Filter low counts rows before HiC-Pro normalization"
},
"ice_filter_high_count_perc": {
"type": "integer",
- "description": "Filter high counts rows before normalization"
+ "description": "Filter high counts rows before HiC-Pro normalization"
},
"ice_eps": {
"type": "number",
"default": 0.1,
- "description": "Threshold for ICE convergence"
+ "description": "Threshold for HiC-Pro ICE convergence"
},
"ice_max_iter": {
"type": "integer",
"default": 100,
- "description": "Maximum number of iteraction for ICE normalization"
+ "description": "Maximum number of iteraction for HiC-Pro ICE normalization"
},
"res_zoomify": {
"type": "string",
@@ -289,10 +293,6 @@
"type": "boolean",
"description": "Do not build contact maps"
},
- "skip_ice": {
- "type": "boolean",
- "description": "Do not run ICE normalization"
- },
"skip_dist_decay": {
"type": "boolean",
"description": "Do not run distance/decay plot"
@@ -517,7 +517,7 @@
"$ref": "#/definitions/valid_pairs_detection"
},
{
- "$ref": "#/definitions/contact_maps_options"
+ "$ref": "#/definitions/contact_maps"
},
{
"$ref": "#/definitions/downstream_analysis"