CLIP-Seq Pipeline

Overview

The CLIP-Seq pipeline (goodwright/clipseq v1.0dev) enables high-resolution studies of RNA binding protein (RBP)-RNA interactions at transcriptomic scale. Flow supports various forms of single-end CLIP data including variants of iCLIP (e.g., irCLIP, iCLIP2, iiCLIP) and eCLIP.

The pipeline is maintained by Goodwright in collaboration with Ule lab and the developers of the nf-core/clipseq pipeline.

Pipeline Summary

The CLIP-Seq pipeline processes your data through these major steps:

1. Quality Control & Preprocessing - Initial FastQC, adapter trimming and quality filtering
2. UMI Extraction - Move UMIs from reads to headers for deduplication
3. Pre-alignment - Map to rRNA/tRNA sequences to filter contaminants
4. Genome Alignment - Align unmapped reads to reference genome with STAR
5. Deduplication - UMI-based deduplication to identify unique crosslinks
6. Peak Calling - Multiple algorithms (Paraclu, Clippy, iCount) for peak detection
7. Motif Analysis - PEKA analysis for enriched k-mer motifs
8. Quality Reports - Comprehensive MultiQC report with CLIP-specific metrics

Input Requirements

Sample Sheet Format

The pipeline requires a CSV sample sheet with the following columns:

Example sample sheet:

sample,fastq_1,fastq_2,group,replicate
WT_rep1,WT_rep1.fastq.gz,,wildtype,1
WT_rep2,WT_rep2.fastq.gz,,wildtype,2
MUT_rep1,MUT_rep1.fastq.gz,,mutant,1
MUT_rep2,MUT_rep2.fastq.gz,,mutant,2

Additional Requirements

• Reference Genome: Soft-masked FASTA file (recommended for PEKA)
• Gene Annotation: GTF file matching the genome version
• Small RNA Reference: FASTA file containing rRNA/tRNA sequences

Key Parameters

Required Parameters

UMI Processing Parameters

Alignment Parameters

Peak Calling Parameters

Output Control Parameters

Pipeline Outputs

Directory Structure

results/
├── 00_genome/          # Reference files and indexes
├── 01_prealign/        # Trimming and QC reports
├── 02_alignment/       # Alignment files
├── 03_filt_dedup/      # Filtered and deduplicated BAMs
├── 04_crosslinks/      # Crosslink positions and coverage
├── 05_peak_calling/    # Peak calls from multiple algorithms
└── 06_reports/         # QC and summary reports

Output Files

• Crosslink Files

  • *.genome.bed: Genomic crosslink positions
  • *.norm.genome.bedgraph: Normalized coverage (CPM) for genome browser
  • *.transcript.bed: Transcriptomic crosslink positions

• Peak Calls

  • clippy/*.clippy.bed: Clippy peak calls
  • paraclu/*.paraclu.peaks.bed: Paraclu clusters
  • icount/*.significant_sites.bed: iCount significant crosslinks

• Analysis Summaries

  • icount/*summary_gene.tsv: Gene-level crosslink counts
  • icount/*summary_type.tsv: RNA type distribution
  • icount/*rnamaps.pdf: RNA maps around landmarks

• Motif Analysis

  • peka/*_enrichment.pdf: PEKA k-mer enrichment plots
  • peka/*_pwm.txt: Position weight matrices

• Quality Reports

  • multiqc/multiqc_report.html: Comprehensive QC report
  • clipqc/*.json: CLIP-specific quality metrics

Example Usage

Basic CLIP-Seq Analysis

nextflow run goodwright/clipseq \
  --samplesheet samples.csv \
  --fasta genome.fa \
  --gtf annotation.gtf \
  --smrna_fasta rRNA_tRNA.fa \
  --outdir results \
  -profile docker

Analysis with UMI in Reads

nextflow run goodwright/clipseq \
  --samplesheet samples.csv \
  --fasta genome.fa \
  --gtf annotation.gtf \
  --smrna_fasta rRNA_tRNA.fa \
  --run_move_umi_to_header true \
  --move_umi "NNNNNN" \
  --outdir results \
  -profile docker

Analysis without UMI Deduplication

nextflow run goodwright/clipseq \
  --samplesheet samples.csv \
  --fasta genome.fa \
  --gtf annotation.gtf \
  --smrna_fasta rRNA_tRNA.fa \
  --run_umi_dedup false \
  --outdir results \
  -profile docker

Tips and Best Practices

Sample Preparation

• Use descriptive sample names: proteinName_cellType_condition_replicate
• Ensure consistent barcode design across experiments
• Include appropriate input/control samples when possible

UMI Handling

• Verify UMI location (in read vs header) for public data
• Check SRA data carefully - headers may be stripped
• Use appropriate UMI pattern (e.g., 'NNNNNN' for 6nt UMI)

Genome Preparation

• Use soft-masked genomes for better PEKA repeat filtering
• Ensure GTF annotation matches genome version
• Include comprehensive rRNA/tRNA sequences for pre-mapping

Parameter Optimization

• Adjust --paraclu_min_value for sparse data or low complexity samples
• Modify --trim_length based on read length distribution
• Tune alignment stringency via --star_params for difficult samples

On Flow Platform

• Run "Prepare CLIP-Seq" first if not already available for your species
• Use the CLIP annotation template for multiplexed samples
• Samples with identical Group and Replicate values will be merged

Troubleshooting

Common Issues

Issue: Low number of unique crosslinks

• Solution: Check UMI handling parameters match library preparation
• Verify adapter trimming is removing all adapter sequences
• Consider adjusting deduplication strategy

Issue: High percentage of reads mapping to rRNA/tRNA

• Solution: Verify library preparation protocol quality
• Check if rRNA depletion was performed
• Ensure comprehensive small RNA reference file

Issue: No peaks detected

• Solution: Lower --paraclu_min_value threshold
• Check crosslink file has sufficient coverage
• Verify correct strand information in analysis

Issue: Pipeline fails at PEKA analysis

• Solution: Ensure genome is soft-masked
• Check for sufficient crosslinks in peak regions
• Try running without PEKA using skip options

Issue: Memory errors during alignment

• Solution: Reduce STAR --limitBAMsortRAM parameter
• Process samples in smaller batches
• Increase memory allocation in config

Additional Resources

• Pipeline Documentation: GitHub - goodwright/clipseq
• nf-core CLIP-seq: nf-co.re/clipseq
• Ule Lab Resources: ulelab.info
• Support: Contact Flow support for pipeline-specific issues
• Citation: Data Science Issues in Studying Protein–RNA Interactions with CLIP Technologies