REAPER: a project-centric workflow layer for comparative repeatome analysis

Abstract

Introduction Repeatome characterization from short-read sequencing data is widely performed using RepeatExplorer2/TAREAN. However, long-lived multisample projects and explicit comparative designs are often executed as ad hoc command sequences that are hard to version, rerun, and monitor on shared compute environments — a gap that motivates a project-centric workflow layer for repeatome analysis. Methods We present REAPER (Repeatome Extended Analysis Pipeline—Execution and Reporting), a project-centric workflow layer that couples a modular Snakemake pipeline with a Python project manager to enforce a stable on-disk layout and configuration-driven execution for single-sample and comparative repeatome analyses. REAPER does not implement a new repeat-discovery algorithm; it is an orchestration layer, and biological accuracy for clustering and satellite calling depends on the underlying RepeatExplorer2/TAREAN and satMiner methods it coordinates. REAPER standardizes: Read QC Deterministic subsampling and preparation RepeatExplorer2/TAREAN execution via seqclust, with satMiner-inspired iterative assembly Post-TAREAN BLAST-based annotation against curated repeat collections (optionally including taxon-scoped NCBI-derived resources with freshness checks) Optional graph-based comparative reports The pipeline makes comparative read allocation, prefix policy, and analysis-ready tables explicit; caching supports incremental reruns and structured logs support monitoring. Performance was assessed using a Triticeae short-read dataset (five samples), with rule-level logging of runtime and memory across pipeline stages. Results Rule-level performance logs show that graph-based clustering dominates runtime and memory, while QC and preparation steps are lightweight by comparison. Graph-report annotations for the Triticeae project additionally link high-ranking clusters to established repeat markers — including pTa794- and pSc119-class entries in curated databases. Discussion These findings illustrate biologically interpretable outputs (recovery of known Triticeae repeat markers) alongside quantitative performance metrics (identification of graph-based clustering as the dominant computational cost). By making comparative read allocation, prefix policy, and analysis-ready tables explicit — and by supporting caching and structured logging — REAPER supports reproducible comparative repeatome analysis in evolving multisample projects. As an orchestration layer rather than a discovery algorithm, REAPER's contribution lies in reproducibility, monitorability, and comparative-analysis infrastructure, with biological accuracy remaining contingent on the underlying RepeatExplorer2/TAREAN and satMiner methods.