Cancer is driven by genomic alterations, called somatic variations. Next generation sequencing (NGS) has revolutionized our ability to determine genomes and compare, for example, tumor to normal cells in order to identify somatic variations. Nonetheless, accurate somatic variation-calling using high-throughput sequence data remains one of the major challenges in cancer genomics. The goal of this project is the design, implementation and evaluation of new methods for somatic variation discovery from NGS reads. Diverse mutation types will be considered. Since this approach is compute-intensive, an efficient implementation on parallel architectures is planned. The developed software will be evaluated using both simulated and real datasets (e.g. from The Cancer Genome Atlas).