Title: Algorithms for ancestry inference on DNA test
Ph.D. Candidate: Yiming Zhang
Major Advisor: Dr. Yufeng Wu
Associate Advisors: Dr. Ion Mandoiu, Dr. Derek Aguiar
Committee Members: Dr. Alexander Russell, Dr. Fei Miao
Date/Time: Tuesday, May 9th, 2023, 10:30 am
Location: HBL1102
Meeting Link:
https://uconn-cmr.webex.com/uconn-cmr/j.php?MTID=m27c1b96895377d652be79170cae50981
Meeting Number:
2623 685 5855
Password:
D22aZPkg5kx
Abstract
Ancestry inference has become a common feature in commercial DNA tests, such as those offered by Ancestry.com and 23andMe. Such tests allow customers to learn about their genetic ancestry by identifying which parts of their genome come from specific ancestral populations. This dissertation focuses on the algorithms used in DNA tests for ancestry inference.
We began by examining a pre-existing algorithm for inferring the admixture proportion of parents from the genotype of an admixed individual. Building on this method, we developed an extended approach using a hidden Markov model (HMM) to jointly infer parental ancestry and genotype calls from the genotype data of a small number of children. Our approach, called parMix, provides fine-scale inference of parental ancestry and genotypes at each single nucleotide polymorphism site.
Next, we present a general approach for inferring the ancestry of an admixed individual’s ancestors. This algorithm estimates the ancestry of all recent ancestors of an extant individual, providing more comprehensive information on their genetic ancestry.
Finally, we propose future work that applies deep learning methods, such as the Transformer, to improve the investigation of ancestry inference.
Joy
Joy Billion
Program Assistant
University of Connecticut
Computer Science and Engineering
371 Fairfield Way, Unit 4155, Room 250
Storrs, CT 06269-4155
Phone (860) 486-3719