Rain in the desert: Bringing machine learning to pulse genetic resources

Authors: Coyne, Clarice - Clare

Submitted to: North American Pulse Improvement Association
Publication Type: Abstract Only
Publication Acceptance Date: 10/27/2021
Publication Date: N/A
Citation: N/A

Interpretive Summary:

Technical Abstract: Plant genetic resource collections are extraordinarily valuable for plant scientists, especially breeders. However, there are significant challenges for efficient use of these collections. For example, many of the USDA cool season food legume accessions were received decades ago, most times with scant passport data and it is unclear how accessions are related genetically or how adapted for a given environment. Fortunately, phenomic and genomic and tools can now be applied to guide plant genetic resource selection for target traits and environments. These are in a nascent state but rapidly gaining traction. Examples will be given for HTPP of 3000 pea accessions (1 environment) and 420 pea accessions (3 environments) for seed protein concentration determination and plant disease resistance in lentil plant genetic resources (2 environments). Genomic advancements are noted with pea, lentil and chickpea core SNP genotyping at 5.8, 4.8 and 4.8% (respectively) of the USDA collection with higher sequencing density and collection coverage planned or in progress to reach 75%. This will enable genomic prediction within the pulse collections. A recent genomic prediction example with the USDA pea core indicates this will be productive for food legume genetic resource selection.