← Về thư mục
📄 / / profiles / r-and-d / raw_evidence / OpenAI4S / skills / bioskills / bio-clip-seq-clip-alignment / references / usage-guide.md

CLIP-seq Alignment - Usage Guide

Overview

Align preprocessed CLIP-seq reads (eCLIP, iCLIP, iCLIP2, PAR-CLIP) to genome with crosslink-preserving parameters. The defining constraints: end-to-end alignment (no 5' soft-clip), strict mismatch ceiling (except PAR-CLIP T->C), and unique-mapper-only by default - with multi-mapper EM rescue (CLAM) reserved for RBPs that bind repetitive elements (Alu, LINE-1, LTR). STAR is the ENCODE standard; HISAT2 is the low-memory alternative; bowtie2 is acceptable only for non-spliced targets.

Prerequisites

conda install -c bioconda star hisat2 bowtie2 samtools umi_tools
# CLAM for multi-mapper EM rescue (repeat-binding RBPs)
pip install CLAM

Quick Start

Tell your AI agent what you want to do: - "Align my eCLIP with the ENCODE STAR parameter block" - "Raise mismatch ceiling for PAR-CLIP to keep T-to-C reads" - "Use HISAT2 because my server only has 16 GB RAM" - "My RBP binds LINE-1 repeats - run STAR with --outFilterMultimapNmax 100 and rescue with CLAM" - "Filter MAPQ >= 10 and UMI-dedupe with method=unique" - "Allele-specific binding study - add WASP filter to STAR"

Example Prompts

ENCODE-style eCLIP alignment

"Align my eCLIP paired-end reads with STAR using --alignEndsType EndToEnd, --outFilterMultimapNmax 1, --outFilterMismatchNoverReadLmax 0.04"

"Sort, index, and UMI-dedupe with umi_tools dedup --method=unique --paired"

PAR-CLIP exception

"Raise STAR mismatch ceiling to 0.07 so the T-to-C signal does not get filtered as sequencing error"

Repeat-binding RBP rescue

"MATR3 binds LINE-1 - emit up to 100 multi-mappers from STAR and feed to CLAM for EM-based reassignment"

"How much does multi-mapper rescue increase my peak count for FUS? Compare unique-only vs CLAM"

Low-memory alternative

"STAR needs 30 GB; my machine has 16 - switch to HISAT2 with --no-softclip"

Allele-specific

"BEAPR-compatible alignment for allele-specific binding - add --waspOutputMode SAMtag with the heterozygous SNP VCF"

Validation

"Confirm my BAM did not soft-clip the truncation site - count reads with S in CIGAR"

"Check that R2 5' positions cluster at expected truncation sites, not at uniform offsets"

What the Agent Will Do

  1. Pick STAR (default), HISAT2 (low memory), or bowtie2 (non-splice target) based on the scenario
  2. Apply the ENCODE eCLIP parameter block: --alignEndsType EndToEnd, --outFilterMultimapNmax 1, --outFilterMismatchNoverReadLmax 0.04, --outFilterType BySJout, --outFilterScoreMinOverLread 0.66, --outFilterMatchNminOverLread 0.66
  3. For PAR-CLIP: override mismatch ceiling to 0.07
  4. For repeat-binding RBPs: raise multi-mapper ceiling to 100 and add CLAM EM rescue
  5. For allele-specific: add WASP filter against het VCF
  6. Index BAM, MAPQ filter (>=10 STAR, >=30 bowtie2/HISAT2), UMI-dedupe with --method=unique
  7. Validate post-alignment: soft-clip rate < 1%, R2 5' positions at expected truncation sites

Tips