Use of Whole Genome Sequence Data to Design a Custom Genotyping Array for American Indians

American Indians have high rates of T2D and obesity, both being heritable characteristics with a strong genetic influence. Yet, genomic studies in this ethnic group have been hampered because currently available genotyping arrays are not well suited for capturing unique genetic architecture or ethnic specific variation. The goal of our study was to thoroughly interrogate the Pima Indian genome for T2D or obesity loci by obtaining whole genome sequence data and designing a custom genotyping array that tags all common variations. Sequence data from 335 Pima Indians identified 13M variants of which ~4.9 M had a minor allele frequency (maf) ≥5% among the sequenced samples and 66,558 variants had a MAF between 1-5% and were potentially functional (altered coding sequence or splice site). Of these 4,986,735 variants (87,850 novel), 4,777,625 (96%) could be captured by 703,797 tag markers (R2>0.85 for redundancy). These tag markers along with 227 mitochondrial markers and 1050 ancestry informative markers were incorporated on a custom genotyping array. To date, 3637 Pima Indians with longitudinal data on T2D and BMI, of which 550 subjects had additional data on related metabolic traits, have been genotyped with this array. Further genotyping in 4060 additional American Indians with longitudinal data on T2D and BMI is ongoing for the “top” 700 variants (variants with lowest p value for association with T2D or BMI in the 3637 subjects). Among those variants currently genotyped in all samples (N=7697), a novel damaging R1420H mutation in ABCC8 (carrier rate of 3.3%) was identified where H-allele carriers had higher birth weights (consistent with higher insulin levels during fetal growth) and an increased risk for T2D (OR=2.02 [1.47-2.78]), and a damaging R611C (rs34052647) in LIPE was identified that associated with T2D (OR= 1.36 [1.18-1.57]) and supported a prior study of LIPE as a diabetes loci in the Old Order Amish.