Unraveling Plant Recombination Patterns: Insights From Genome k-mers

Area of Science:

• Genetics and Genomics
• Bioinformatics
• Plant Breeding

Background:

• Crossover recombination during meiosis generates genetic diversity, crucial for plant breeding and gene transfer.
• Predicting recombination events is vital for efficiently transferring desirable gene variants between plant varieties.
• Genome structure features are increasingly used to understand and predict recombination.

Purpose of the Study:

• To investigate the relationship between genome structure, quantified by k-mers, and crossover recombination rates.
• To introduce the kmerExtractor Python package for k-mer analysis from genome data.
• To develop and evaluate machine learning models for predicting recombination rates using k-mer features.

Main Methods:

• Utilized the kmerExtractor package employing frequency chaos game representation (FCGR) to count k-mers from genome fasta files.
• Analyzed k-mer frequencies across six plant species: Arabidopsis, bean, maize, rice, sorghum, and tomato.
• Trained regression-based machine learning models using k-mer derived features to predict recombination rates.

Main Results:

• Identified both positive and negative correlations between specific k-mers (3-mers, 2-mers) and recombination rates.
• Demonstrated the predictive power of k-mer information for recombination rates, especially in sorghum and tomato.
• Observed linear relationships between certain k-mers and recombination events in predictive models.

Conclusions:

• K-mer analysis provides valuable genomic sequence information for understanding and predicting crossover recombination.
• The developed predictive strategy shows promise for guiding new plant crosses and accelerating breeding programs.
• This approach can enhance the efficiency of transferring beneficial traits in crop improvement.