Benchmarking small-variant genotyping in polyploids

Area of Science:

• Genomics
• Plant Breeding
• Bioinformatics

Background:

• Genotyping from sequencing is crucial for polyploid plant molecular breeding.
• Existing benchmarks for genotyping accuracy are limited for polyploids, unlike diploids.
• Accurate genotyping in polyploids is essential for advancing breeding strategies.

Purpose of the Study:

• To evaluate genotyping error rates for polyploid plants using real sequencing data.
• To benchmark the performance of the Octopus variant calling method against other popular tools.
• To establish benchmarks for assessing genotyping accuracy in polyploid variant calling.

Main Methods:

• In silico mixtures of diploid Genome in a Bottle (GIAB) samples were used to create tetraploid and hexaploid datasets.
• Whole-genome sequencing data from these simulated polyploids were analyzed.
• The Octopus variant calling method was evaluated alongside other popular genotyping tools.
• Concordance analysis was performed on real autotriploid banana datasets.

Main Results:

• Genotyping errors are prevalent in polyploid datasets at typical sequencing depths.
• Octopus demonstrated a 25% reduction in genotyping errors on average compared to other methods.
• Benchmarks were established for evaluating genotyping accuracy in polyploid variant calling.

Conclusions:

• Octopus offers improved accuracy for genotyping small variants in polyploid plants.
• The developed benchmarks are vital for assessing and improving polyploid genotyping tools.
• This work facilitates more reliable molecular breeding in polyploid species.