extraChIPs

Workflows

Standard Workflow

This package builds on existing tools and adds some simple but extremely useful capabilities for working wth ChIP-Seq data. The focus is on detecting differential binding windows/regions. One set of functions focusses on set-operations retaining mcols for GRanges objects, whilst another group of functions are to aid visualisation of results. Coercion to tibble objects is also implemented.

library(extraChIPs)
library(GenomicRanges)

# Define chromosome information
sq <- defineSeqinfo("GRCh37")

# Create mock GRangesList for consensus peak calling
gr1 <- GRanges("chr10:1000-2000", seqinfo = sq)
gr2 <- GRanges("chr10:1200-2200", seqinfo = sq)
peaks <- GRangesList(sample1 = gr1, sample2 = gr2)

# Generate consensus peaks
consensus <- makeConsensus(peaks, p = 0.5)

# Plot overlaps
plotOverlaps(peaks)

Input: A GRangesList of peak calls; Output: A consensus GRanges object and overlap plots.

When to Use

• Consensus Peak Definition: To define consensus peaks across replicates using makeConsensus.
• Peak Importing and Filtering: To import and filter peaks with blacklists/greylists using importPeaks.
• Differential Signal Analysis: To perform differential signal analysis on window/region counts using fitAssayDiff.
• Gene Mapping: To map peaks/regions to genes and promoters using mapByFeature.
• Visualization: To visualize peak overlaps with plotOverlaps and signal profiles with plotProfileHeatmap or getProfileData.

When NOT to Use

• De Novo Motif Discovery: For de novo motif discovery, use memes or universalmotif instead.
• De Novo Peak Calling: For peak calling itself, use MACS2 or epigraHMM as extraChIPs is designed for downstream analysis of existing peak calls.

Data Requirements

• Peak Files: Peak files in narrowPeak format or GRanges objects.
• Read Counts: BAM files for read counting, or a pre-computed RangedSummarizedExperiment object.
• Annotations: Annotation files (GTF/GFF) for mapping peaks to genes.

Key Parameters

• p (0.5): Minimum proportion of replicates in which a peak must be present to be included in the consensus.
• norm ("TMM"): Normalization method in fitAssayDiff (e.g., "TMM" or library-size).
• fc (1.2): Fold-change threshold incorporated into testing in fitAssayDiff.
• asRanges (TRUE): Logical indicating whether to return results as a GRanges object.
• upstream (2500): Upstream distance for defining promoters.
• downstream (500): Downstream distance for defining promoters.

Best Practices

• Define Seqinfo: Define a consistent Seqinfo object at the start of the workflow using defineSeqinfo.
• Exclude Artifacts: Exclude blacklisted and grey-listed regions using importPeaks to avoid false positives.
• Normalization Check: Test whether group-specific count distributions are similar using quantro before applying TMM normalization.
• Classify Status: Use addDiffStatus to classify regions as "Increased", "Decreased", or "Unchanged" for downstream visualization.

Common Pitfalls

• Inappropriate Normalization: Applying TMM normalization when group-specific distributions differ significantly; use quantro to test this assumption first.
• Metadata Loss: Losing metadata columns during GRanges set operations; use reduceMC or makeConsensus to retain metadata.
• Mapping Coordinates: Mapping peaks to genes without resetting the core ranges; use colToRanges to restore the original peak boundaries before mapping.

Alternatives

• DiffBind: For standard affinity-based differential analysis.
• csaw: For sliding window-based differential binding analysis.
• ChIPseeker: For peak annotation and visualization.

Citations

• Ross-Innes et al. 2012, Nature (for DiffBind-style approaches)
• Hicks and Irizarry 2015, Genome Biol. (for quantro)

References

• Homepage: bioconductor.org/packages/extraChIPs
• Vignette: https://bioconductor.org/packages/release/bioc/vignettes/extraChIPs/inst/doc/extraChIPs.html