A Neural Network Model for Maximizing Prediction Accuracy in Haplotype Tagging SNP Selection

Due to the tremendous number of single nucleotide polymorphisms (SNPs), there is a clear need to expedite genotyping by considering only a subset of all SNPs called haplotype tagging SNPs (htSNPs). Recently, the approach that selects htSNPs by maximizing their prediction accuracy has demonstrated very promising results. Here we propose a new prediction system for htSNP selection based on neural network models. We applied our system to three public data sets, and compared its prediction performance to that of two state-of-the-art prediction rules. The results demonstrate that our system consistently outperforms compared methods with robust performance.