VarCA

VarCA identifies single nucleotide variants (SNVs), insertions/deletions (indels), and de novo mutations in regulatory regions by analyzing ATAC-seq reads to enable discovery of non-coding regulatory variants without relying on whole-genome sequencing (WGS).

Key Features:

• Ensemble integration: Integrates features from seven distinct variant callers using an ensemble approach to predict variants more accurately than individual callers.
• Caller set: Applies seven distinct variant callers, with explicit evaluation of individual caller performance including the Genome Analysis Toolkit (GATK).
• Data types: Processes both bulk and single-cell ATAC-seq read data for variant detection.
• Variant classes: Detects and evaluates single nucleotide variants (SNVs), insertions/deletions (indels), and de novo mutations in regulatory regions.
• Peak-region focus: Operates on ATAC-seq peaks and reports performance within peak regions containing at least ten reads.
• Reported individual performance: Identifies GATK as the best-performing individual caller for SNVs (precision/recall 0.92/0.97 in bulk ATAC-seq peak regions ≥10 reads) and for indels (precision/recall 0.87/0.82).
• Reported ensemble performance: Reports ensemble precision/recall of 0.99/0.95 for SNVs and 0.93/0.80 for indels on bulk ATAC-seq, and 0.98/0.94 for SNVs and 0.82/0.82 for indels on single-cell ATAC-seq.
• ATAC-seq rationale: Leverages ATAC-seq reads from regulatory sequences as an alternative to WGS for capturing regulatory variants.

Scientific Applications:

• Regulatory variant discovery: Identification of non-coding regulatory variants from ATAC-seq data.
• De novo mutation detection: Detection of de novo mutations within regulatory regions using ATAC-seq reads.
• Variant caller benchmarking: Comparative evaluation and benchmarking of variant callers on bulk and single-cell ATAC-seq datasets, including assessment of GATK performance.
• Single-cell variant calling: Calling SNVs and indels from single-cell ATAC-seq to enable variant analysis at single-cell resolution.

Methodology:

VarCA applies seven distinct variant callers to bulk and single-cell ATAC-seq reads, extracts features from individual callers and integrates them via an ensemble approach to predict SNVs and indels, and evaluates performance using precision and recall within ATAC-seq peak regions (≥10 reads), reporting GATK as the top individual caller.