Predicting body mass index and waist-hip ratio from genome-wide single nucleotide polymorphism data.

Research output: Contribution to conferencePaperpeer-review


Obesity has become important health concern in the last few decades, especially in developed countries. Body mass index (BMI) a measure of overall adiposity, is the traditional index of obesity. However, waist-hip ratio (WHR) a measure of abdominal adiposity is increasingly being used as an alternative index of obesity. In this study, best linear unbiased prediction with a genomic relationship matrix (G-BLUP) derived from 263,357 genome-wide SNPs, was used to predict unobserved phenotypes of the obesity indices body mass index and WHR within 2,159 Croatian and into 805 UK individuals. Predictive accuracy was evaluated using 10 fold cross validation. Heritability estimates (95% confidence intervals [CI]) were 0.28(0.14-0.42) and 0.58(0.36-0.71) for BMI, and 0.35(0.21-0.49) and 0.48 (0.30-0.66) for HWR in the Croatian and UK data respectively. The prediction accuracies (95% CI) obtained for the two obesity indices were similar when predicting within the Croatian data; 0.10(0.10-0.11) for BMI and 0.10 (0.10-0.11) for WHR. However when predicting into the UK data; a higher prediction accuracy of 0.08 (0.082-0.083) was achieved for BMI, as compared to 0.02(0.020-0.021) for WHR. The disparity observed in the prediction accuracies of BMI and WHR in the UK data may relate to differences in the trait architectures of the two obesity indices. Future work will involve fitting models which can accommodate trait architectures that depart from the infinitesimal model assumed by G-BLUP.
Original languageEnglish
Publication statusPublished - 30 May 2014
EventQuantitative Genomics 2014 - Academy of Medical Sciences, London, United Kingdom
Duration: 30 May 201430 May 2014


ConferenceQuantitative Genomics 2014
Country/TerritoryUnited Kingdom


Dive into the research topics of 'Predicting body mass index and waist-hip ratio from genome-wide single nucleotide polymorphism data.'. Together they form a unique fingerprint.

Cite this