eCLIP is a pipeline designed to identify genomic locations of RNA-bound proteins.

Please refer to the script that contains prerequisites for this pipeline: source create_environment_clipseq.sh

To get you started, this provides you with a reference data file and a tutorial .dataset

These files contain everything needed to run a small example

(make sure to place this in a location with plenty of space!):

• (IP sample) Read 1 FASTQ.gz
• (IP sample) Read 2 FASTQ.gz (not applicable for single-end)
• (size-matched input sample) Read 1 FASTQ.gz
• (size-matched input sample) Read 2 FASTQ.gz (not applicable for single-end)
• (chromosome 19 only) STAR index directory
• (repbase) STAR index directory
• (barcodes) FASTA file containing barcodes for demultiplexing reads (for single-end, use "a_adapters.fasta")
• (chrom sizes) chrom.sizes file (tabbed file containing chromosome name and length, can be downloaded from UCSC)

Execute the analysis using the provided example YAML file pointing to the appropriate bundled example files

cd example/
cwl-runner ../cwl/wf_get_peaks_scatter_pe.cwl paired_end_clip_small.yaml (paired end)
cwl-runner ../cwl/wf_get_peaks_scatter_se.cwl single_end_clip_small.yaml (single end *warning*: no small dataset available, just using r1 of paired-end for now as a small example...)

• At the top of each YAML file, there will be either:
  • eCLIP_pairedend: run paired-end pipeline using TOIL batch runner (default)
  • eCLIP_singleend: run single-end pipeline using TOIL batch runner (default)
  • wf_get_peaks_scatter_pe.cwl: run paired-end pipeline using cwl reference runner
  • wf_get_peaks_scatter_se.cwl: run single-end pipeline using cwl reference runner

You can change these depending on your need. TOIL batch runner will be set by default to submit to a TORQUE cluster with at least 16 cores per node, and 64Gb memory per node. If you are unsure, best to run this pipeline using the cwl-runner script that should be installed via create_environment_clipseq.sh (see example above)

You may run into libstc++ errors. See various online forums for more info, as solutions will vary by machine.

Running time for the examples should be ~15 minutes. Running on a complete dataset takes about a day for human ENCODE data (24 hours), so sit back and relax by reading the rest of this README.

These are the minimum required arguments needed to run the pipeline (you can view the same information inside the wf_get_peaks.yaml file):

dataset: kbp550  # name prefixed onto outputs

I have a command that removes temporary directories (on success) that start with those prefixes (rm -rf out_tmp*)

Add STAR directories:

speciesGenomeDir:
  class: Directory
  path: /path/to/stargenome

repeatElementGenomeDir:
  class: Directory
  path: /path/to/repeatelement

CLIPPER params:

species: hg19  # for supported species, see clipper docs

UMI & barcode params:

randomer_length: "5"  # (Paired-end only) length of the UMI assigned to each read

barcodesfasta:  # (Paired-end only) This is a FASTA formatted file containing the barcodes we will use to demultiplex our FASTQ's:
  class: File
  path: /path/to/barcodes

a_adapters:  # (Single-end only) This is a processed Ril19 set of sequences to be trimmed from SE reads
  class: File
  path: /path/to/a_adapters.fasta

The following YAML block describes the location paths of the forward (read1), reverse (read2) reads, and the barcodes required to demultiplex these reads for each sample.

Barcode names must match those described in the above barcodes.fasta file!

(For example, if you are using our standard paired-end barcodes https://github.com/YeoLab/eclip/blob/master/example/inputs/yeolabbarcodes_20170101.fasta, make sure the barcodeids are one of: A01, A03, A04, B06, C01, D8f, F05, G07, X1A, X1B, X2A, X2B, or NIL for "inputs". Single-end protocols do not have inline barcodes, and you will use the a_adapters.fasta instead.

We're showing two samples (2 replicates each) for a paired-end experiment described in this space. Each sample will be defined as indicated below each name: field. Make sure these names are unique per sample! They (and dataset name above) are used to determine the filename prefixes.

samples:
  -
    - ip_read:
      name: rep1_clip
      barcodeids: [A01, B06]  # remove this line if processing single-end data
      read1:
        class: File
        path: /path/to/clip.fastq.gz
      read2:  # remove this line if processing single-end data
        class: File  # remove this line if processing single-end data
        path: /path/to/clip.fastq.gz  # remove this line if processing single-end data

    - input_read:
      name: rep1_input
      barcodeids: [NIL, NIL]
      read1:
        class: File
        path: /path/to/clip.fastq.gz
      read2:
        class: File
        path: /path/to/clip.fastq.gz
  -
    - ip_read:
      name: rep2_clip
      barcodeids: [C01, D8f]
      read1:
        class: File
        path: /path/to/clip.fastq.gz
      read2:
        class: File
        path: /path/to/clip.fastq.gz

    - input_read:
      name: rep2_input
      barcodeids: [NIL, NIL]
      read1:
        class: File
        path: /path/to/clip.fastq.gz
      read2:
        class: File
        path: /path/to/clip.fastq.gz

For Single-end experiments, the samples format will look slightly different (although other fields will be the same):

samples:
  - 
    - ip_read:
      name: rep1_clip
      read1:
        class: File
        path: /path/to/fastq.gz

    - input_read:
      name: rep1_input
      read1:
        class: File
        path: /path/to/fastq.gz
  - 
    - ip_read:
      name: rep2_clip
      read1:
        class: File
        path: /path/to/fastq.gz

    - input_read:
      name: rep2_input
      read1:
        class: File
        path: /path/to/fastq.gz

Input-normalized peaks will contain regions of binding.

made with: https://www.tablesgenerator.com/markdown_tables

• When going through the merged BAM file results, I can only find files with only one of the paired barcodes (e.g. A01 of A01/B06). Is this normal? Yes, *.merged*.bam indicates that both barcodes have been merged, I just use the first as a prefix namespace for the next step.

Van Nostrand, Eric L., et al. "Robust, Cost-Effective Profiling of RNA Binding Protein Targets with Single-end Enhanced Crosslinking and Immunoprecipitation (seCLIP)." mRNA Processing. Humana Press, New York, NY, 2017. 177-200.

Van Nostrand, E.L., Pratt, G.A., Shishkin, A.A., Gelboin-Burkhart, C., Fang, M.Y., Sundararaman, B., Blue, S.M., Nguyen, T.B., Surka, C., Elkins, K. and Stanton, R. "Robust transcriptome-wide discovery of RNA-binding protein binding sites with enhanced CLIP (eCLIP)." Nature methods 13.6 (2016): 508-514.

Amstutz, Peter; Crusoe, Michael R.; Tijanić, Nebojša; Chapman, Brad; Chilton, John; Heuer, Michael; Kartashov, Andrey; Leehr, Dan; Ménager, Hervé; Nedeljkovich, Maya; Scales, Matt; Soiland-Reyes, Stian; Stojanovic, Luka (2016): Common Workflow Language, v1.0. figshare. https://doi.org/10.6084/m9.figshare.3115156.v2 Retrieved: 22 13, May 11, 2017 (GMT)

Kurtzer GM, Sochat V, Bauer MW (2017): Singularity: Scientific containers for mobility of compute. PLoS ONE 12(5): e0177459. https://doi.org/10.1371/journal.pone.0177459