bio-workflows-chipseq-pipeline

What this skill does

## Version Compatibility

Reference examples tested with: Bowtie2 2.5.3+, MACS3 3.0+, HOMER 4.11+, bedtools 2.31+, fastp 0.23+, samtools 1.19+

Before using code patterns, verify installed versions match. If versions differ:
- R: `packageVersion('<pkg>')` then `?function_name` to verify parameters
- CLI: `<tool> --version` then `<tool> --help` to confirm flags

If code throws ImportError, AttributeError, or TypeError, introspect the installed
package and adapt the example to match the actual API rather than retrying.

# ChIP-seq Pipeline

**"Process my ChIP-seq data from FASTQ to annotated peaks"** -> Orchestrate QC, Bowtie2 alignment, duplicate removal, MACS3 peak calling, ChIPseeker annotation, and QC metrics (FRiP, strand cross-correlation).

Complete workflow from raw ChIP-seq FASTQ files to annotated peaks.

## Workflow Overview

```
FASTQ files (IP + Input)
    |
    v
[1. QC & Trimming] -----> fastp
    |
    v
[2. Alignment] ---------> Bowtie2
    |
    v
[3. BAM Processing] ----> sort, markdup, filter
    |
    v
[4. Peak Calling] ------> MACS3
    |
    v
[5. QC] ----------------> FRiP, fingerprint plots
    |
    v
[6. Annotation] --------> ChIPseeker
    |
    v
Annotated peaks + QC report
```

## Primary Path: Bowtie2 + MACS3 + ChIPseeker

### Step 1: Quality Control with fastp

```bash
# Process both IP and Input samples
for sample in IP_rep1 IP_rep2 Input_rep1 Input_rep2; do
    fastp -i ${sample}_R1.fastq.gz -I ${sample}_R2.fastq.gz \
        -o trimmed/${sample}_R1.fq.gz -O trimmed/${sample}_R2.fq.gz \
        --detect_adapter_for_pe \
        --qualified_quality_phred 20 \
        --length_required 25 \
        --html qc/${sample}_fastp.html
done
```

### Step 2: Alignment with Bowtie2

```bash
# Build index (once)
bowtie2-build genome.fa bt2_index/genome

# Align
for sample in IP_rep1 IP_rep2 Input_rep1 Input_rep2; do
    bowtie2 -p 8 -x bt2_index/genome \
        -1 trimmed/${sample}_R1.fq.gz \
        -2 trimmed/${sample}_R2.fq.gz \
        --no-mixed --no-discordant \
        --maxins 1000 \
        2> aligned/${sample}.log | \
    samtools view -@ 4 -bS -q 30 - | \
    samtools sort -@ 4 -o aligned/${sample}.bam
done
```

**QC Checkpoint:** Check alignment rate
- Overall alignment >80%
- Unique mapping >70%

### Step 3: BAM Processing

```bash
for sample in IP_rep1 IP_rep2 Input_rep1 Input_rep2; do
    # Mark and remove duplicates
    samtools fixmate -m aligned/${sample}.bam - | \
    samtools sort - | \
    samtools markdup -r - aligned/${sample}.dedup.bam

    # Index
    samtools index aligned/${sample}.dedup.bam

    # Remove chrM reads (high mitochondrial is common)
    samtools view -h aligned/${sample}.dedup.bam | \
        grep -v chrM | \
        samtools view -b - > aligned/${sample}.final.bam
    samtools index aligned/${sample}.final.bam
done
```

### Step 4: Peak Calling with MACS3

```bash
# Narrow peaks (TFs, sharp histone marks like H3K4me3)
macs3 callpeak \
    -t aligned/IP_rep1.final.bam aligned/IP_rep2.final.bam \
    -c aligned/Input_rep1.final.bam aligned/Input_rep2.final.bam \
    -f BAMPE \
    -g hs \
    -n experiment \
    --outdir peaks \
    -q 0.01

# Broad peaks (H3K27me3, H3K36me3)
macs3 callpeak \
    -t aligned/IP_rep1.final.bam aligned/IP_rep2.final.bam \
    -c aligned/Input_rep1.final.bam aligned/Input_rep2.final.bam \
    -f BAMPE \
    -g hs \
    -n experiment_broad \
    --outdir peaks \
    --broad \
    --broad-cutoff 0.1
```

For higher-confidence peaks, run HOMER as well and intersect results (recommended for final peak sets). When using `--nomodel`, estimate fragment size from cross-correlation or `macs3 predictd` rather than using a generic default; 147bp (nucleosome core) is the biologically grounded fallback for histone marks. For HOMER, use `-style histone` for all histone marks including H3K4me3. See chip-seq/peak-calling for HOMER commands and multi-caller consensus guidance.

### Step 5: QC Metrics

```bash
# Calculate FRiP (Fraction of Reads in Peaks)
total_reads=$(samtools view -c aligned/IP_rep1.final.bam)
reads_in_peaks=$(bedtools intersect -a aligned/IP_rep1.final.bam -b peaks/experiment_peaks.narrowPeak -u | samtools view -c)
frip=$(echo "scale=4; $reads_in_peaks / $total_reads" | bc)
echo "FRiP: $frip"

# Generate bigWig for visualization
bamCoverage -b aligned/IP_rep1.final.bam \
    -o bigwig/IP_rep1.bw \
    --normalizeUsing RPKM \
    -p 8

# Fingerprint plot (assess enrichment)
plotFingerprint \
    -b aligned/IP_rep1.final.bam aligned/Input_rep1.final.bam \
    --labels IP Input \
    -o qc/fingerprint.pdf
```

**QC Checkpoint:** Assess enrichment quality
- FRiP >1% (ideally >5% for good enrichment)
- Fingerprint shows clear separation between IP and Input

### Step 6: Peak Annotation

When a custom GTF is provided, use it directly via `makeTxDbFromGFF()` (R), `annotatePeaks.pl -gtf` (HOMER), or Python. See chip-seq/peak-annotation for all three approaches. Only fall back to pre-built TxDb packages (e.g., `TxDb.Hsapiens.UCSC.hg38.knownGene`) when no project-specific annotation is available.

```r
library(ChIPseeker)
library(GenomicFeatures)
library(rtracklayer)

# Custom GTF approach (preferred when a GTF is provided)
txdb <- makeTxDbFromGFF('annotation.gtf', format = 'gtf')
# Standard genome approach (when no custom GTF)
# library(TxDb.Hsapiens.UCSC.hg38.knownGene)
# txdb <- TxDb.Hsapiens.UCSC.hg38.knownGene

peaks <- readPeakFile('peaks/experiment_peaks.narrowPeak')
# overlap='all' couples gene assignment with feature overlap (host-gene convention);
# default overlap='TSS' assigns nearest-TSS gene independently of feature overlap
peak_anno <- annotatePeak(peaks, TxDb = txdb, tssRegion = c(-2000, 2000), overlap = 'all')

# Map gene symbols from GTF (annoDb only works with pre-built TxDb)
gtf <- import('annotation.gtf')
gene_map <- unique(data.frame(
    gene_id = sub('\\..*', '', gtf$gene_id),
    symbol = gtf$gene_name, stringsAsFactors = FALSE))
anno_df <- as.data.frame(peak_anno)
anno_df$geneId_base <- sub('\\..*', '', anno_df$geneId)
anno_df$SYMBOL <- gene_map$symbol[match(anno_df$geneId_base, gene_map$gene_id)]

plotAnnoPie(peak_anno)
plotDistToTSS(peak_anno)

write.csv(anno_df, 'peaks/annotated_peaks.csv', row.names = FALSE)
promoter_genes <- unique(anno_df$SYMBOL[grepl('Promoter', anno_df$annotation)])
write.table(promoter_genes, 'peaks/promoter_genes.txt', row.names = FALSE, col.names = FALSE, quote = FALSE)
```

## Parameter Recommendations

| Step | Parameter | Narrow Peaks | Broad Peaks |
|------|-----------|--------------|-------------|
| MACS3 | --broad | No | Yes |
| MACS3 | -q | 0.01 | - |
| MACS3 | --broad-cutoff | - | 0.1 |
| MACS3 | -g | hs/mm/ce/dm | Same |
| Bowtie2 | -q (samtools) | 30 | 30 |

## Troubleshooting

| Issue | Likely Cause | Solution |
|-------|--------------|----------|
| Few peaks | Low enrichment, wrong parameters | Check fingerprint, adjust -q threshold |
| Many peaks | High noise, PCR duplicates | Remove duplicates, use stricter -q |
| Low FRiP | Poor antibody, low enrichment | Check antibody, increase sequencing |
| Peaks in blacklist | Technical artifacts | Filter against ENCODE blacklist |

## Complete Pipeline Script

```bash
#!/bin/bash
set -e

THREADS=8
GENOME="genome.fa"
INDEX="bt2_index/genome"
IP_SAMPLES="IP_rep1 IP_rep2"
INPUT_SAMPLES="Input_rep1 Input_rep2"
OUTDIR="results"

mkdir -p ${OUTDIR}/{trimmed,aligned,peaks,qc,bigwig}

# Step 1: QC
for sample in $IP_SAMPLES $INPUT_SAMPLES; do
    fastp -i ${sample}_R1.fastq.gz -I ${sample}_R2.fastq.gz \
        -o ${OUTDIR}/trimmed/${sample}_R1.fq.gz \
        -O ${OUTDIR}/trimmed/${sample}_R2.fq.gz \
        --html ${OUTDIR}/qc/${sample}_fastp.html -w ${THREADS}
done

# Step 2-3: Align and process
for sample in $IP_SAMPLES $INPUT_SAMPLES; do
    bowtie2 -p ${THREADS} -x ${INDEX} \
        -1 ${OUTDIR}/trimmed/${sample}_R1.fq.gz \
        -2 ${OUTDIR}/trimmed/${sample}_R2.fq.gz \
        --no-mixed --no-discordant 2> ${OUTDIR}/qc/${sample}_align.log | \
    samtools