Unraveling Plant Recombination Patterns: Insights From Genome k-mers

Crossover recombination is a pivotal event that takes place during meiosis of germinal cells, leading to the rearrangement of parental chromosomes and generating novel allele combinations, thereby enhancing genetic diversity. This process holds significant importance for plant breeders as it enables the transfer of gene variants from one variety to another. Recent studies have explored diverse strategies to predict recombination events along chromosomes in key plant species, employing various types of genome features. In this study, the relationship between genome structure, quantified using k-mers, and crossover recombination is investigated. To facilitate this analysis, the Python package kmerExtractor is introduced; it uses frequency chaos game representation (FCGR) to count k-mers from genome fasta files and adds them as column features for subsequent analysis. This package is used to explore the genomes of one model and five crop plant species, namely, Arabidopsis, bean, maize, rice, sorghum, and tomato. The investigation reveals both positive and negative trends between 3-mers, 2-mers, and recombination rates. Furthermore, the information derived from k-mers was used to train regression-based machine learning models for predicting recombination rates along chromosomes. The results demonstrate the efficacy of using k-mer for predicting purposes, particularly for sorghum and tomato datasets, highlighting linear relationships between several k-mers and recombination events. We hope that this predictive strategy based on genomic sequence information can be useful for the development of new plant crosses.