PIPE-53-psf-to-bedpe (#148)

ENCODE-DCC · Mar 8, 2022 · 83f4fbf · 83f4fbf
1 parent 787615b
commit 83f4fbf
Show file tree

Hide file tree

Showing 5 changed files with 60 additions and 22 deletions.
diff --git a/docker/hic-pipeline/Dockerfile b/docker/hic-pipeline/Dockerfile
@@ -102,10 +102,6 @@ RUN git clone https://github.com/ENCODE-DCC/kentUtils_bin_v381.git && \
     cd ../../ && \
     rm -rf kentUtils_bin_v381
 
-RUN git clone --branch phasing https://github.com/aidenlab/3d-dna.git && \
-    cd 3d-dna && \
-    git checkout 634b49b9a93ac166914dc3baae586489bb793c92
-
 # Install Juicer
 RUN git clone --branch encode https://github.com/theaidenlab/juicer.git && \
     cd juicer && \
@@ -150,6 +146,14 @@ RUN git clone https://github.com/aidenlab/hic2gatk.git && \
     cd hic2gatk && \
     git checkout dee2a9f6f2b0e95f0cb3e8d47eef9798cb8101aa
 
+RUN git clone --branch phasing https://github.com/aidenlab/3d-dna.git && \
+    cd 3d-dna && \
+    git checkout 63029aa3bc5ba9bbdad9dd9771ace583cc95e273
+
+RUN git clone https://github.com/sa501428/psf-to-bedpe.git && \
+    cd psf-to-bedpe && \
+    git checkout 0.1
+
 # For sorting, LC_ALL is C
 ENV LC_ALL C
 ENV PATH=/opt:/opt/scripts:/opt/scripts/common:/opt/juicer/misc:/opt/gatk-4.2.2.0:/opt/hic2gatk:$PATH

diff --git a/genophase.wdl b/genophase.wdl
@@ -4,9 +4,9 @@ import "./hic.wdl"
 
 workflow genophase {
     meta {
-        version: "1.11.3"
-        caper_docker: "encodedcc/hic-pipeline:1.11.3"
-        caper_singularity: "docker://encodedcc/hic-pipeline:1.11.3"
+        version: "1.12.0"
+        caper_docker: "encodedcc/hic-pipeline:1.12.0"
+        caper_singularity: "docker://encodedcc/hic-pipeline:1.12.0"
         croo_out_def: "https://raw.githubusercontent.com/ENCODE-DCC/hic-pipeline/dev/croo_out_def.json"
     }
 
@@ -25,8 +25,8 @@ workflow genophase {
         Int? run_3d_dna_ram_gb
         Boolean no_phasing = false
 
-        String docker = "encodedcc/hic-pipeline:1.11.3"
-        String singularity = "docker://encodedcc/hic-pipeline:1.11.3"
+        String docker = "encodedcc/hic-pipeline:1.12.0"
+        String singularity = "docker://encodedcc/hic-pipeline:1.12.0"
     }
 
     RuntimeEnvironment runtime_environment = {
@@ -73,6 +73,11 @@ workflow genophase {
             disk_size_gb = run_3d_dna_disk_size_gb,
             runtime_environment = runtime_environment,
         }
+
+        call convert_psf_to_bedpe { input:
+            psf = run_3d_dna.psf,
+            runtime_environment = runtime_environment,
+        }
     }
 }
 
@@ -228,6 +233,35 @@ task run_3d_dna {
 
         File assembly_in = "snp.out.in.assembly.gz"
         File assembly = "snp.out.out.assembly.gz"
+
+        File psf = "out.psf"
+    }
+
+    runtime {
+        cpu : "~{num_cpus}"
+        disks: "local-disk ~{disk_size_gb} HDD"
+        memory: "~{ram_gb} GB"
+        docker: runtime_environment.docker
+        singularity: runtime_environment.singularity
+    }
+}
+
+task convert_psf_to_bedpe {
+    input {
+        File psf
+        Int num_cpus = 1
+        Int disk_size_gb = 1000
+        Int ram_gb = 16
+        RuntimeEnvironment runtime_environment
+    }
+
+    command <<<
+        set -euo pipefail
+        awk -f /opt/psf-to-bedpe/psf-to-bedpe.awk ~{psf} > "psf.bedpe"
+    >>>
+
+    output {
+        File bedpe = "psf.bedpe"
     }
 
     runtime {

diff --git a/hic.wdl b/hic.wdl
@@ -19,9 +19,9 @@ struct RuntimeEnvironment {
 
 workflow hic {
     meta {
-        version: "1.11.3"
-        caper_docker: "encodedcc/hic-pipeline:1.11.3"
-        caper_singularity: "docker://encodedcc/hic-pipeline:1.11.3"
+        version: "1.12.0"
+        caper_docker: "encodedcc/hic-pipeline:1.12.0"
+        caper_singularity: "docker://encodedcc/hic-pipeline:1.12.0"
         croo_out_def: "https://raw.githubusercontent.com/ENCODE-DCC/hic-pipeline/dev/croo_out_def.json"
         description: "ENCODE Hi-C pipeline, see https://github.com/ENCODE-DCC/hic-pipeline for details."
     }
@@ -65,10 +65,10 @@ workflow hic {
         Int? create_accessibility_track_disk_size_gb
         String assembly_name = "undefined"
 
-        String docker = "encodedcc/hic-pipeline:1.11.3"
-        String singularity = "docker://encodedcc/hic-pipeline:1.11.3"
-        String delta_docker = "encodedcc/hic-pipeline:1.11.3_delta"
-        String hiccups_docker = "encodedcc/hic-pipeline:1.11.3_hiccups"
+        String docker = "encodedcc/hic-pipeline:1.12.0"
+        String singularity = "docker://encodedcc/hic-pipeline:1.12.0"
+        String delta_docker = "encodedcc/hic-pipeline:1.12.0_delta"
+        String hiccups_docker = "encodedcc/hic-pipeline:1.12.0_hiccups"
     }
 
     RuntimeEnvironment runtime_environment = {

diff --git a/hic_pipeline/__init__.py b/hic_pipeline/__init__.py
@@ -1,5 +1,5 @@
 __title__ = "hic-pipeline"
-__version__ = "1.11.3"
+__version__ = "1.12.0"
 __description__ = "ENCODE Hi-C uniform processing pipeline."
 __url__ = "https://github.com/ENCODE-DCC/hic-pipeline"
 __uri__ = __url__

diff --git a/make_restriction_site_locations.wdl b/make_restriction_site_locations.wdl
@@ -7,9 +7,9 @@ struct RuntimeEnvironment {
 
 workflow make_restriction_site_locations {
     meta {
-        version: "1.11.3"
-        caper_docker: "encodedcc/hic-pipeline:1.11.3"
-        caper_singularity: "docker://encodedcc/hic-pipeline:1.11.3"
+        version: "1.12.0"
+        caper_docker: "encodedcc/hic-pipeline:1.12.0"
+        caper_singularity: "docker://encodedcc/hic-pipeline:1.12.0"
     }
 
     parameter_meta {
@@ -22,8 +22,8 @@ workflow make_restriction_site_locations {
         File reference_fasta
         String assembly_name
         String restriction_enzyme
-        String docker = "encodedcc/hic-pipeline:1.11.3"
-        String singularity = "docker://encodedcc/hic-pipeline:1.11.3"
+        String docker = "encodedcc/hic-pipeline:1.12.0"
+        String singularity = "docker://encodedcc/hic-pipeline:1.12.0"
     }