On marker-based parentage verification via non-linear optimization

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• Parentage verification traditionally relies on short tandem repeat (STR) markers.
• Single nucleotide polymorphisms (SNPs) are increasingly preferred for genotyping due to their bi-allelic nature.
• Existing SNP-based parentage verification algorithms have limitations, such as failing to exclude all non-parents.

Purpose of the Study:

• To develop and evaluate a novel algorithm for parentage verification using constrained regression.
• To overcome limitations of existing methods in rejecting non-parental candidates.
• To assess the algorithm's speed and accuracy with varying numbers of SNPs.

Main Methods:

• Developed a parentage verification algorithm based on constrained regression.
• Tested the algorithm on 14,816 animals using 50, 100, and 500 SNP genotypes.
• Evaluated parentage assignment using coefficient ranking and a minimum coefficient threshold.
• Assessed performance using the power of assignment (P[Formula: see text]) and power of exclusion (P[Formula: see text]).

Main Results:

• The algorithm achieved high power of assignment and exclusion (both >0.98) with 50 SNPs when the true sire was present.
• Using a coefficient threshold, the power of assignment remained high (>0.99), while the power of exclusion was robust across SNP numbers.
• The algorithm demonstrated strong performance in rejecting non-parents when the true sire was absent, with powers of exclusion up to 1.

Conclusions:

• The constrained regression algorithm is a fast, accurate, and easily implementable method for parentage verification.
• It effectively utilizes genomic marker data, even with a minimal set of 50 SNPs.
• This approach offers a significant improvement for SNP-based parentage assignment in various populations.