Communications in Information and Systems

Volume 10 (2010)

Number 2

Prediction of Phenotype Information from Genotype Data

Pages: 99 – 114



Jens Gramm

Richard M. Karp

William S. Noble

Roded Sharan

Qian-fei Wang

Nir Yosef


The dissection of complex diseases is one of the greatest challenges of human genetics with important clinical and scientific applications. Traditionally, associations were sought between single genetic markers and disease. The availability of large scale SNP data makes it possible, for the first time, to study the predictive power of genotypes and haplotypes with respect to phenotype data. Here we present a novel method for predicting phenotype information from genotype data. The method is based on a support vector machine that employs new kernel functions for the similarity between genotypes or their underlying haplotypes. We demonstrate our approach on SNP data for the apolipoprotein gene cluster in baboons, predicting plasma lipid levels with significant success rates, and identifying associations that were not detected using extant approaches.


Machine learning (Computing Methodologies-Artificial Intelligence-Learning); Parameter learning (Computing Methodologies-Artificial Intelligence-Learning); Classifier design and evaluation (Computing Methodologies-Pattern Recognition-Design Methodology); Biology and genetics (Computer Applications-Life and Medical Sciences)

Full Text (PDF format)