High performance genomic sequencing: A filtered approach

Abstract

Protein and DNA homology detection systems are an essential part in computational biology applications. These algorithms have changed over the time from dynamic programming approaches by finding the optimal local alignment between two sequences to statistical approaches with different kinds of heuristics that minimize former executions times. However, the continuously increasing size of input datasets is being projected into the use of High Performance Computing (HPC) hardware and software in order to address this problem. The aim of the research presented in this paper is to propose a new filtering methodology, based on general-purpose graphical processor units (GP-GPUs) and multi-core processors, for removing those sequences considered irrelevant in terms of homology and similarity. The proposed methodology is completely independent from the homology detection algorithm. This approach is very useful for researchers and practitioners because they do not need to understand a new algorithm. This design has been approved by the National Biotechnology Research Center of Spain (CNB).