Use of overlapping DNA pools to discern genetic differences despite pooling error

Authors: Keele, John; McDaneld, Tara; Kuehn, Larry

Submitted to: Journal of Animal Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/22/2023
Publication Date: 5/25/2023
Citation: Keele, J.W., McDaneld, T.G., Kuehn, L.A. 2023. Use of overlapping DNA pools to discern genetic differences despite pooling error. Journal of Animal Science. 101. Article skad166. https://doi.org/10.1093/jas/skad166.
DOI: https://doi.org/10.1093/jas/skad166

Interpretive Summary: Genotyping arrays are widely used in genetic evaluation programs to increase accuracy of genetic selection. However, genotyping all animals with useful phenotypes can be cost prohibitive; for example, animals from commercial settings can add relevant data, but they will not enter genetic selection programs themselves. Pooling of DNA is a cost saving method to genotype groups of animals to use their data in genetic evaluation. However, assay measurement error or pool construction errors may decrease the benefit of these pools. This research demonstrates the use of overlapping pools to quantify variation due to these errors as a case study to use this technique in larger projects to increase DNA pooling utility and accuracy. DNA pooling provides a cost-effective approach to capturing these data and genetically linking seedstock and commercial populations.

Technical Abstract: Genotyping pools of commercial cattle and individual seedstock animals may reveal hidden relationships between sectors enabling use of commercial data for genetic evaluation. However, commercial data capture may be compromised by inexact pool formation. We aimed to estimate the concordance between distances or genomic covariance among pooling allele frequencies (PAFs) of DNA pools comprised of 100 animals with 0% or 50% overlap of animals in common between pools. Cattle lung samples were collected from a commercial beef processing plant on a single day. Six pools of 100 animals each were constructed so that overlap between pools was 0% or 50%. Two pools of all 200 animals were constructed to estimate PAFs for all 200 animals. Frozen lung tissue (0.01 g) from each animal was weighed into a tube containing a pool; there were two pools of 200 animals each and six pools of 100 animals each. Every contribution of an individual animal was an independent measurement to insure independence of pooling errors. Lung samples were kept on dried ice during the pooling process to keep them from thawing. The eight pools were then assayed for approximately 100,000 single nucleotide polymorphisms (SNP). PAF for each SNP and pool was based on the relative intensity of the two dyes used to detect the alleles rather than genotype calls which are not tractable from pooling data. Euclidean distances and genomic relationships among the PAFs for the eight pools were estimated and distances were tested for concordance with pool overlap using permutation-based analysis of distance. Distances among pools were concordant with the planned overlap of animals shared between pools (P = 0.0024); pool overlap accounted for 70% of the variation and pooling error accounted for 30%. Pools containing 100 animals with no overlap were the most distant from one another and pools with 50% overlap were the least distant. This work shows that we can discern differences in distance between pairs of overlapping DNA pools sharing 0% and 50% of the animals. Genomic correlations among nonoverlapping pools indicated that nonoverlapping pool pairs did not share many related animals because genomic correlations were near zero for these pairs. On the other hand, one pair of nonoverlapping pools likely contained related animals between pools because the correlation was 0.21. Pools sharing 50% overlap ranged in genomic relationship between 0.21 and 0.39 (N = 12).