TY - JOUR
T1 - Predictive DNA analysis for biogeographical ancestry
AU - Cheung, Elaine Y.Y.
AU - Gahan, Michelle Elizabeth
AU - McNevin, Dennis
PY - 2018/11/2
Y1 - 2018/11/2
N2 - Establishment of national DNA databases in Australia and overseas has increased the number of criminal convictions, yet a high volume of serious crime cases remain with no suspect profile nor any DNA database matches. In these circumstances prediction of biogeographical ancestry (BGA) and externally visible characteristics can assist by providing forensic intelligence in conjunction with, or in place of, eyewitness testimonies. To predict the BGA of an individual requires: genetic markers selected for their ability to differentiate between BGAs; representative BGA reference populations; and a prediction algorithm (‘classifier’) that predicts the BGA of an unknown individual based on genetic markers in the reference populations. The human genome contains autosomal ancestry informative markers that are easily harvested from publicly accessible collections of genotypes with associated ancestry information. A number of classification methods are available including Bayesian approaches and distance-based algorithms. BGA is likely to be continuous rather than discrete and some methods are inappropriate for the prediction of admixed BGA. As predictive services become available to the public and private sectors, there is a risk of results being misinterpreted if an inappropriate tool is applied. Understanding the underlying marker sets, reference populations and classification algorithms is required to prevent ill-informed predictions.
AB - Establishment of national DNA databases in Australia and overseas has increased the number of criminal convictions, yet a high volume of serious crime cases remain with no suspect profile nor any DNA database matches. In these circumstances prediction of biogeographical ancestry (BGA) and externally visible characteristics can assist by providing forensic intelligence in conjunction with, or in place of, eyewitness testimonies. To predict the BGA of an individual requires: genetic markers selected for their ability to differentiate between BGAs; representative BGA reference populations; and a prediction algorithm (‘classifier’) that predicts the BGA of an unknown individual based on genetic markers in the reference populations. The human genome contains autosomal ancestry informative markers that are easily harvested from publicly accessible collections of genotypes with associated ancestry information. A number of classification methods are available including Bayesian approaches and distance-based algorithms. BGA is likely to be continuous rather than discrete and some methods are inappropriate for the prediction of admixed BGA. As predictive services become available to the public and private sectors, there is a risk of results being misinterpreted if an inappropriate tool is applied. Understanding the underlying marker sets, reference populations and classification algorithms is required to prevent ill-informed predictions.
KW - ancestry informative markers
KW - Biogeographical ancestry
KW - classification
KW - genetic distance
KW - multidimensional scaling
KW - STRUCTURE
UR - http://www.scopus.com/inward/record.url?scp=85040998035&partnerID=8YFLogxK
UR - http://www.mendeley.com/research/predictive-dna-analysis-biogeographical-ancestry
U2 - 10.1080/00450618.2017.1422021
DO - 10.1080/00450618.2017.1422021
M3 - Article
AN - SCOPUS:85040998035
SN - 0045-0618
VL - 50
SP - 651
EP - 658
JO - Australian Journal of Forensic Sciences
JF - Australian Journal of Forensic Sciences
IS - 6
ER -