Skip to main content

Accurate and Reliable Prediction of HLA Class II Alleles Using SNPs in an African Population

Thursday, October 11, 2012 — Poster Session III

10:00 a.m. – Noon

Natcher Conference Center, Building 45




  • F Tekola Ayele
  • E Hailu
  • C Finan
  • A Aseffa
  • G Davey
  • MJ Newport
  • CN Rotimi
  • A Adeyemo


The human leukocyte antigen (HLA) genes play important roles in the immune system and have multiple alleles that vary widely across human populations. However, direct HLA typing is time consuming and too expensive. Despite reports of high prediction performance of in silico HLA-prediction methods in European populations, they have not been evaluated in African population datasets. In the present study, we predicted HLA-DRB1 and DQB1 alleles using SNP data in an Ethiopian population with aim to provide simple, inexpensive, and highly accurate methods for African populations. The subjects comprised 297 Ethiopians with genome-wide SNP data, of whom 188 were HLA typed. Prediction was performed using the Multi-allelic Gene Prediction (MAGPrediction) method. Accuracy of the prediction was evaluated by comparing predicted and directly typed HLA alleles, and discriminative ability of the model’s prediction. We found that the model predicted two-digit resolution for HLA-DRB1 and DQB1 alleles at accuracy levels of 96% and 87%, respectively. All measures of performance showed high accuracy and reliability of the prediction. In conclusion, this study showed that HLA alleles can be predicted with high accuracy in an African population using SNP genotype data. These findings offer relatively less expensive opportunities for HLA studies in African populations.

back to top