Large-scale comparative genomics poses computational challenges both in runtime and accuracy. My lab has been developing algorithms and software for comparing genome sequences and for predicting disease risk and phenotype from genomic data. In this talk I will present our work on predicting chronic lymphocytic leukemia using biomarkers that we discover in exome sequences of 180 cases and 155 controls obtained from the NIH. I will also discuss our new GPU program for mapping divergent short reads to a genome and present results obtained with it on real data.