High-throughput genotype (HTG) platforms allow the simultaneous screening of many single nucleotide polymorphisms (SNP) across many genes per sample for a reasonable price. This analysis presents an algorithm for the implementation of HTG data in nonlinear mixed effect (NLME) modeling techniques, including model evaluation. The suitability of the developed algorithm is tested by applying it to four simulated data sets representing various genotype–phenotype relationships.
An algorithm consisting of repeated preselection of SNPs by analysis of variance (ANOVA) and integration of these SNPs into the NLME model by a forward inclusion/backward elimination procedure was applied to four simulated pharmacokinetic datasets with 300 patients each, 1200 SNPs per patient and various genotype-clearance relationships. Analysis was performed using SAS 9.1.3 and NONMEM VI.
The algorithm discovered all true positive genotype–phenotype relationships. In two datasets, no false positive (FP) relationships were incorporated into the model, whereas in the remaining two datasets, three FP relationships were revealed. By application to independent datasets and subsequent backward elimination, the FP rate was reduced to zero in all four datasets.
An algorithm is presented as a complementary method to existing data analysis methods. It allows the successful integration of HTG data into NLME models for the detection of genotype–phenotype relationships. Independent evaluation datasets were valuable tools to reduce the number of FP findings. Overall, this algorithm allows the advantages of NLME modeling and HTG data to be combined synergistically and might help to provide new insights into the genotype–phenotype relationship of drugs.