kalis

kalis provides a high-performance implementation of the Li & Stephens hidden Markov model for local ancestry inference, enabling rapid resolution of ancestry at loci of interest through posterior decodings represented as N x N distance matrices.

Key Features:

• High-performance computation: Exploits multi-core parallelism and modern CPU vector instruction sets to scale to hundreds of thousands of genomes.
• Minimal pre-processing: Makes posterior decodings readily accessible with minimal pre-processing requirements.
• Extensible framework: Enables developers to build a range of variant-specific ancestral inference pipelines on top of the core engine.
• Distance matrix output: Produces N x N distance matrices based on posterior decodings for downstream analyses.

Scientific Applications:

• Local ancestry inference: Resolves the recent ancestry at specific loci of interest in phased haplotype data.
• Selection studies: Enables genome-wide screens for selection signals through local ancestry analysis.
• Association studies: Supports genome-wide association studies by providing local ancestry information.
• Introgression detection: Facilitates detection of introgression signals via local ancestry inference.
• Phylogeny approximation: Approximates the recent phylogeny of N phased haplotypes at a set of variants along the genome.

Methodology:

Implements the Li & Stephens hidden Markov model for inferring recent ancestry at a given variant, computing posterior decodings represented as N x N distance matrices.