PIP-1663-add-loop-localizer (#136)

.pre-commit-config.yaml

@@ -18,13 +18,7 @@ repos:
       - id: isort
         language_version: python3.7
 
-  - repo: git://github.com/detailyang/pre-commit-shell
-    sha: v1.0.6
-    hooks:
-      - id: shell-lint
-        args: [--severity=warning]
-
-  - repo: git://github.com/detailyang/pre-commit-shell
+  - repo: https://github.com/detailyang/pre-commit-shell
     sha: v1.0.6
     hooks:
       - id: shell-lint

docker/hic-pipeline/Dockerfile

@@ -116,6 +116,10 @@ RUN curl \
     chmod 666 /opt/juicer/CPU/common/juicer_tools.jar && \
     ln -s juicer/CPU scripts
 
+RUN curl \
+        -LO \
+        https://github.com/aidenlab/Juicebox/releases/download/v.2.14.00/feature_tools.jar
+
 RUN curl \
         -LO \
         https://github.com/sa501428/mixer-tools/releases/download/v4.07.09/MixerTools.4.7.9.jar&& \

docker/hiccups/Dockerfile

@@ -40,7 +40,7 @@ RUN curl \
 
 RUN curl \
         -LO \
-        https://github.com/aidenlab/Juicebox/releases/download/v.2.13.07/feature_tools.jar
+        https://github.com/aidenlab/Juicebox/releases/download/v.2.14.00/feature_tools.jar
 
 # For sorting, LC_ALL is C
 ENV LC_ALL C

hic.wdl

@@ -14,9 +14,9 @@ struct BamAndLigationCount {
 
 workflow hic {
     meta {
-        version: "1.8.0"
-        caper_docker: "encodedcc/hic-pipeline:1.8.0"
-        caper_singularity: "docker://encodedcc/hic-pipeline:1.8.0"
+        version: "1.9.0"
+        caper_docker: "encodedcc/hic-pipeline:1.9.0"
+        caper_singularity: "docker://encodedcc/hic-pipeline:1.9.0"
         croo_out_def: "https://raw.githubusercontent.com/ENCODE-DCC/hic-pipeline/dev/croo_out_def.json"
     }
 
@@ -34,8 +34,8 @@ workflow hic {
 
         # Parameters controlling delta calls
         Boolean no_delta = false
-        String delta_docker = "encodedcc/hic-pipeline:1.8.0_delta"
-        String hiccups_docker = "encodedcc/hic-pipeline:1.8.0_hiccups"
+        String delta_docker = "encodedcc/hic-pipeline:1.9.0_delta"
+        String hiccups_docker = "encodedcc/hic-pipeline:1.9.0_hiccups"
 
         Boolean intact = false
         Array[String] normalization_methods = []
@@ -250,12 +250,19 @@ workflow hic {
                     docker = hiccups_docker,
                 }
             }
+
             if (intact) {
                 call hiccups_2 { input:
                     hic = add_norm.output_hic,
                     quality = qualities[i],
                     docker = hiccups_docker,
                 }
+
+                call localizer as localizer_intact { input:
+                    hic = add_norm.output_hic,
+                    loops = hiccups_2.merged_loops,
+                    quality = qualities[i],
+                }
             }
         }
 
@@ -283,6 +290,11 @@ workflow hic {
             resolutions = delta_resolutions,
             models_path = delta_models_path,
         }
+
+        call localizer as localizer_delta { input:
+            hic = if length(add_norm.output_hic) > 1 then add_norm.output_hic[1] else select_first([input_hic]),
+            loops = delta.loops,
+        }
     }
 
     if (defined(input_hic)) {
@@ -298,11 +310,16 @@ workflow hic {
                     docker = hiccups_docker,
                 }
             }
+
             if (intact) {
                 call hiccups_2 as hiccups_2_input_hic { input:
                     hic = select_first([input_hic]),
                     docker = hiccups_docker,
                 }
+                call localizer as localizer_intact_input_hic { input:
+                    hic = select_first([input_hic]),
+                    loops = hiccups_2_input_hic.merged_loops,
+                }
             }
         }
     }
@@ -912,6 +929,47 @@ task hiccups_2 {
     }
 }
 
+task localizer {
+    input {
+        File hic
+        File loops
+        Int quality = 0
+        Int num_cpus = 24
+    }
+
+    command {
+        set -euo pipefail
+        export LOOPS_FILE=loops.bedpe
+        gzip -dc ~{loops} > $LOOPS_FILE
+        java \
+            -Ddevelopment=false \
+            -Djava.awt.headless=true \
+            -Xmx60G \
+            -Xms60G \
+            -jar /opt/feature_tools.jar \
+            localizer \
+            -k SCALE \
+            -r 100 \
+            -w 1 \
+            --threads ~{num_cpus} \
+            ~{hic} \
+            $LOOPS_FILE \
+            localized
+        gzip -n localized/localizedList_primary_100.bedpe
+        mv localized/localizedList_primary_100.bedpe.gz localized_loops_~{quality}.bedpe.gz
+    }
+
+    output {
+        File localized_loops = "localized_loops_~{quality}.bedpe.gz"
+    }
+
+    runtime {
+        cpu : "~{num_cpus}"
+        disks: "local-disk 100 HDD"
+        memory: "64 GB"
+    }
+}
+
 task delta {
     input {
         File hic

hic_pipeline/__init__.py

@@ -1,5 +1,5 @@
 __title__ = "hic-pipeline"
-__version__ = "1.8.0"
+__version__ = "1.9.0"
 __description__ = "ENCODE Hi-C uniform processing pipeline."
 __url__ = "https://github.com/ENCODE-DCC/hic-pipeline"
 __uri__ = __url__

make_restriction_site_locations.wdl

@@ -2,9 +2,9 @@ version 1.0
 
 workflow make_restriction_site_locations {
     meta {
-        version: "1.8.0"
-        caper_docker: "encodedcc/hic-pipeline:1.8.0"
-        caper_singularity: "docker://encodedcc/hic-pipeline:1.8.0"
+        version: "1.9.0"
+        caper_docker: "encodedcc/hic-pipeline:1.9.0"
+        caper_singularity: "docker://encodedcc/hic-pipeline:1.9.0"
     }
 
     parameter_meta {

megamap.wdl

@@ -4,9 +4,9 @@ import "./hic.wdl"
 
 workflow megamap {
     meta {
-        version: "1.8.0"
-        caper_docker: "encodedcc/hic-pipeline:1.8.0"
-        caper_singularity: "docker://encodedcc/hic-pipeline:1.8.0"
+        version: "1.9.0"
+        caper_docker: "encodedcc/hic-pipeline:1.9.0"
+        caper_singularity: "docker://encodedcc/hic-pipeline:1.9.0"
         croo_out_def: "https://raw.githubusercontent.com/ENCODE-DCC/hic-pipeline/dev/croo_out_def.json"
     }