Run ENCODE ATAC-seq Pipeline on the Whitehead Server

If you have human (hg38, hg19) and mouse (mm10, mm9) samples with biological replicates, you run ​ENCODE ATAC-seq Pipeline. The pipeline takes fastq files, cleans and maps the reads, filters aligned reads and does peak calls. Here is the ​schema of the workflow. In addition, it does quality controls. Here is a ​ sample QC report. The steps below shows you how to run it on our Whitehead server. Note: It only works on python2.

• content in input sample.json:

  {
      "atac.pipeline_type" : "atac",
      "atac.genome_tsv" : "/nfs/BaRC_datasets/ENCODE_ATAC-seq_Pipeline/mm10/mm10.tsv",
      "atac.fastqs_rep1_R1" : [
          "/fullpath/sample_rep1_1.fastq.gz"
      ],
      "atac.fastqs_rep1_R2" : [
      	"/fullpath/sample_rep1_2.fastq.gz"
      ],
      "atac.fastqs_rep2_R1" : [
      	"/fullpath/sample_rep2_1.fastq.gz"
      ],
      "atac.fastqs_rep2_R2" : [
  	"/fullpath/sample_rep2_2.fastq.gz"
      ],
      "atac.paired_end" : true,
      "atac.auto_detect_adapter" : true,
      "atac.enable_tss_enrich" : true,
      "atac.title" : "sample",
      "atac.description" : "ATAC-seq mouse sample"
  }
  

• Supported genome files for hg19, hg38, mm9 and mm10 can be found in /nfs/BaRC_datasets/ENCODE_ATAC-seq_Pipeline, and atac.genome_tsv used for .json is
  • hg19: /nfs/BaRC_datasets/ENCODE_ATAC-seq_Pipeline/hg19/hg19.tsv
  • hg38: /nfs/BaRC_datasets/ENCODE_ATAC-seq_Pipeline/hg38/hg38.tsv
  • mm9: /nfs/BaRC_datasets/ENCODE_ATAC-seq_Pipeline/mm9/mm9.tsv
  • mm10: /nfs/BaRC_datasets/ENCODE_ATAC-seq_Pipeline/mm10/mm10.tsv

• To initiate conda inside Whitehead:

  # Be sure to keep the first dot in the command below:
  . /nfs/BaRC_Public/conda/start_barc_conda
  

• Before running the ENCODE pipeline, verify there is no preexisting conda startup code with the command below:

  conda env list
  

  You have no preexisting conda if you get "conda: command not found". Otherwise, log out, log back in, start the new conda instance, and activate encode-atac-seq-pipeline
• Ignore the developer's instructions and use your home directory for conda and the pipeline.

  conda activate encode-atac-seq-pipeline
  

• Run. Files could be url or fullpath. ​ Detailed information about .json file

  caper run /nfs/BaRC_Public/atac-seq-pipeline/atac.wdl -i sample.json
  # After the job finishes, you can deactivate conda with
  conda deactivate
  

• The QC report is call-qc_report/execution/qc.html
• idr peaks files:
  • rep1: call-idr_pr/shard-0/execution/rep1-pr1_vs_rep1-pr2.idr0.05.bfilt.narrowPeak.gz
  • rep2: call-idr_pr/shard-1/execution/rep2-pr1_vs_rep2-pr2.idr0.05.bfilt.narrowPeak.gz
  • Note: shard-0 refers to the first biological replicate, shard-1 refers to the 2nd biological replicate, and so on
  • rep1 and rep2: call-idr/shard-1/execution/rep1_vs_rep2.idr0.05.bfilt.narrowPeak.gz