Computational strategies for eukaryotic pangenome analyses

Abstract

Over the last few years, pangenome analyses have been applied to eukaryotes, especially to important crops. A handful of eukaryotic pangenome studies have demonstrated widespread variation in gene presence/absence among plant species and its implications on agronomically important traits. In this chapter, we focus on the methodology of pangenome analysis, which can generally be classified into two different types of approaches, a homolog-based strategy and a "map-to-pan" strategy. In a homolog-based strategy, the genomes of individuals are independently assembled, and the presence/absence of a gene family is determined by clustering protein sequences into homologs. Alternatively, in a "map-to-pan" strategy, pangenome sequences are constructed by combining a well-annotated reference genome with newly identified non-reference representative sequences, from which the presence/absence of a gene is then determined based on read coverage after individual reads are mapped to the pangenome. We highlight the advantages and limitations of the homolog-based strategy and several variant approaches to the "map-to-pan" strategy. We conclude that the "map-to-pan" strategy is highly recommended for eukaryotic pangenome analysis. However, programs and parameters for pangenome analysis need to be carefully selected for eukaryotes with different genome sizes.