Variant calling in non-model organisms with snpArcher

Population genomic studies in non-model organisms increasingly depend on whole-genome resequencing, yet translating raw reads into reliable variant callsets remains a practical challenge due to the complexity of multi-step bioinformatics pipelines and the absence of species-specific best practices. Here we present a step-by-step protocol for snpArcher, a Snakemake-based workflow that takes raw sequencing reads and a reference genome as input and produces a filtered, joint-called VCF suitable for downstream population genomic analysis. We guide users through six phases: installation and environment setup, sample sheet creation, run configuration, execution on local or high-performance computing systems, quality control review using an interactive HTML dashboard, and downstream analysis, focusing on postprocessing and filtering. The QC dashboard aggregates individual-level metrics including principal component analysis, relatedness estimation, depth-missingness diagnostics, and admixture analysis to help identify batch effects, contamination, cryptic relatedness, and outlier samples before downstream analysis. We demonstrate the impact of sequential filtering steps on the site frequency spectrum and demographic inference using a dataset of 137 burrowing owl (Athene cunicularia) genomes, showing how removal of low-coverage individuals, sex-linked scaffolds, and regions of excess heterozygosity eliminates artifacts that would otherwise bias inference of population size history. This protocol is intended as a practical companion to the original snpArcher publication, enabling researchers working with non-model organisms to produce and evaluate analysis-ready variant callsets in a reproducible manner.