Update preprocessing workflow

paulsengroup · Mar 7, 2024 · e9bf42a · e9bf42a
1 parent 4a72149
commit e9bf42a
Show file tree

Hide file tree

Showing 2 changed files with 42 additions and 0 deletions.
diff --git a/preprocessing/configs/preprocess_data.config b/preprocessing/configs/preprocess_data.config
@@ -29,6 +29,10 @@ process {
         container = 'ghcr.io/paulsengroup/2023-hictk-paper/hic-tools:3.30.00'
         memory = 750.GB
     }
+    withName:pairs_to_single_res_hic9 {
+        container = 'ghcr.io/paulsengroup/2023-hictk-paper/hic-tools:3.30.00'
+        memory = 750.GB
+    }
     withName:hictk_convert {
         container = 'ghcr.io/paulsengroup/2023-hictk-paper/hictk-bench:1.0.0'
     }

diff --git a/preprocessing/workflows/preprocess_data.nf b/preprocessing/workflows/preprocess_data.nf
@@ -18,6 +18,12 @@ workflow {
         pairs
     )
 
+    pairs_to_single_res_hic9(
+        process_chrom_sizes.out.chrom_sizes,
+        prepare_pairs_for_juicer.out.txt.collect(),
+        Channel.fromList(params.resolutions)
+    )
+
     pairs_to_hic8(
         process_chrom_sizes.out.chrom_sizes,
         prepare_pairs_for_juicer.out.txt,
@@ -214,6 +220,38 @@ process cooler_zoomify {
         '''
 }
 
+
+process pairs_to_single_res_hic9 {
+    publishDir params.data_dir, mode: 'copy'
+
+    label 'process_medium'
+    label 'process_very_long'
+
+    tag "${pairs.simpleName}"
+
+    input:
+        path chrom_sizes
+        path pairs
+        val resolution
+
+    output:
+        path "*.hic9", emit: hic
+
+    shell:
+        memory_gb=task.memory.toGiga()
+        dest="${pairs.simpleName}.${resolution}.hic9"
+        '''
+        java -Xmx!{memory_gb}G -Xms!{memory_gb}G -jar "$HICTOOLS_JAR" \\
+            pre '!{pairs}'             \\
+                '!{dest}'              \\
+                '!{chrom_sizes}'       \\
+                -j !{task.cpus}        \\
+                --threads !{task.cpus} \\
+                -r '!{resolution}'     \\
+                -n
+        '''
+}
+
 process hictk_convert {
     publishDir "${params.data_dir}/input/matrices", mode: 'copy'