#39 · Tools & infrastructure

Graph representations of genetic variation scale across ARG, genotyping, and simulation

The common thread is scalable graph-based representation of genetic variation, applied at three layers. tskit 1.0 is announced as a stable library for ancestral recombination graphs with documented stability guarantees, positioning population-scale ARG operations as infrastructure. COSIGT applies a pangenome graph approach to genotyping, inferring structural genotypes from low-coverage short reads by matching coverage patterns to pangenome haplotypes via cosine similarity, and is described as population-scalable. GrgPhenoSim extends the genotype representation graph framework to phenotype simulation, scaling to biobank-size datasets and reported dozens to hundreds of times faster. What is new is the pairing of stability commitments with throughput claims across representation, inference, and simulation layers. These speed and scaling figures appear only in the item summaries.