SimFFPE

SimFFPE simulates artifact chimeric reads (ACRs) and normal reads from formalin-fixed paraffin-embedded (FFPE) next-generation sequencing (NGS) data to support evaluation and mitigation of false-positive structural variant (SV) calls.

Key Features:

• R package simulation: Implements read simulation as an R package capable of generating artifact chimeric reads and normal reads from FFPE-derived sequencing data.
• Scope of simulation: Simulates reads across whole genomes, specific chromosomes, or large genomic regions to model diverse genomic contexts.
• Artifact chimeric read generation: Produces mixtures of characteristic ACRs and normal reads derived from combinations of single-stranded DNA fragments with short reverse complementary regions (SRCRs).
• Companion filtration: Works with the complementary tool FilterFFPE, which removes artifact chimeric reads while preserving genuine chimeric reads to reduce false-positive SV calls without substantial loss of sensitivity.
• Performance assessment: Has been evaluated with SV callers Delly, Lumpy, and Manta, demonstrating improved mean positive predictive value (PPV) from 0.27 to 0.48 on simulated samples and from 0.11 to 0.27 on real samples after filtration.

Scientific Applications:

• Structural variant detection accuracy: Facilitates assessment and improvement of SV calling accuracy in FFPE-derived NGS data by modeling and removing ACR-induced false positives.
• Impact analysis of FFPE artifacts: Enables study of how ACRs from single-stranded DNA fragments with SRCRs affect split-read support and SV calling.
• Benchmarking and validation: Supports validation and benchmarking of SV callers and filtration strategies using simulated and real FFPE datasets, with quantified PPV improvements.

Methodology:

Generates controlled mixtures of simulated reads that reflect FFPE artifact profiles and applies an algorithmic filtration process (FilterFFPE) to distinguish artifact chimeric reads from genuine chimeric reads.