Unveiling Genomic Complexity: A Framework for Genome Graph Structural Analysis and Optimised Variant Calling Workflows

Genome graphs offer a powerful alternative to linear reference genomes, as they provide a richer representation of a collection of genomes by emphasising the polymorphic regions. Despite their advantages, it is difficult to characterise the structural complexity of a genome graph and efficiently call variants using it. In our study, we developed a novel framework to visualise and quantify the structural complexity of genome graphs and optimised existing genome graph-based variant-calling workflows. We applied these techniques on two human pan-genome graphs constructed with the variants from the 1000 Genomes Project - the first augmenting only common variants and the second all the variants, including rare ones. We found that rare variants increased the variability of a genome graph by 10-fold but hypervariability by 50-fold. Our framework highlighted biologically significant and highly variable loci like HLA and DEFB and several novel variable regions. Benchmarking the optimised variant-calling workflow with the two genome graphs identified 9.83% more variants than the linear reference genome. Additionally, it showed that rare variants had minimal impact on variant-calling performance. Our methods are effective in identifying novel genomic loci and variants from genome graphs and revealing the underlying genomic complexity of the represented population or species.