Chromatin Structure
Juicer
GCATemplates available: no
Juicer is a pipeline for generating Hi-C maps from fastq raw data files and command line tools for feature annotation on the Hi-C maps.
Hi-C is a genome-wide Chromatin Conformation Capture protocol using proximity ligation.
CPU version
The CPU version is for running on a single GPU compute node.
The first steps (bwa, dedup, merge, hic files) are run on the compute node's CPU and the postproc step is run on the GPU.
-
Load the CPU version of Juicer and set up your directories before submitting your job script
module spider Juicer/1.6-Java-1.9.0.4-CUDA-8.0.61-CPU
-
Copy the Juicer scripts directory to your working directory.
cp -r $EBROOTJUICER/CPU/scripts/ ./
-
Create a directory in your working directory named fastq that contains all your fastq files.
mkdir fastq
-
Then you will need references and restriction_sites directories in your working directory.
There are some references available which you can use by doing the following:
ln -s /scratch/data/bio/Juicer/references ln -s /scratch/data/bio/Juicer/restriction_sites
-
You job script should be in the same directory as your scripts/ fastq/ references/ and restriction_sites/ directories
By default, Juicer will use 48 threads for bwa so configure the #SBATCH parameters for 48 threads unless you want to specify fewer with the jucier.sh -t option
#!/bin/bash #SBATCH --export=NONE # do not export current env to the job #SBATCH --job-name=juicer_cpu # set job name #SBATCH --time=1-00:00:00 # set the wall clock limit to 1 day #SBATCH --nodes=1 # request one node #SBATCH --cpus-per-task=48 # Each task uses 28 cores #SBATCH --mem=360G # request 360 total job memory #SBATCH --gres=gpu:2 # request 1 GPU per node with 2 GPUs #SBATCH --partition=gpu # request the GPU partition/queue #SBATCH --output=stdout.%x.%j #SBATCH --error=stderr.%x.%j module load Juicer/1.6-Java-1.9.0.4-CUDA-8.0.61-CPU ./scripts/juicer.sh