Evaluating Robustness of Algorithm for Microsatellite Marker Genotyping

Abstract

Microsatellites provide powerful genetic tools for complex disease mapping. Microsatellite genotyping requires analyzing peak data for discrimination of the true allele. In a previous study, we developed a new algorithm for automated genotyping. Here, we evaluate our algorithm’s robustness. First, we found that our algorithm calculates the model parameter of noise peaks appropriately and infers genotypes correctly even with low selectivity and specificity in the intermediate result of its first step. Our results indicate the model robustly calculates noise peaks. Second, our algorithm adequately infers true allele peaks for small sample sets. Furthermore, we evaluated its potential risk of failing to construct noise peak model.