Parallel Guided Dynamic Programming Approach for DNA Sequence Similarity Search

Abstract

 Development of DNA sequence comparison technique is an active research activity in computational biology application.Commonly techniques studied are dynamic programming and heuristic algorithms. Exhaustive dynamic programming algorithm produces optimal result but requires longer time and bigger space. Heuristic algorithm gives approximate results with much faster processing. We have developed a new model that improves the speed of large scale DNA sequence similarity search and at the same time the best possible alignment result is retained. The model is known as a guided dynamic programming approach for DNA sequence similarity search (FRA-Search). Two approaches are used to complete the FRA-Search model: an automaton based exact string matching algorithm is employed to skip irrelevant database sequences from being computed for dynamic programming alignment processing and; the rough sets theory has been employed to classify and reduct the dataset. This paper discusses the parallel model for FRA-Search application. The parallel FRA-Search model is implemented on PC-based cluster system. It is developed on a single program multiple data (SPMD) architecture and MPJ Express software is used as a communication interface protocol between processors.

A distributed system for fast alignment of next-generation sequencing data

“…In our work, the distributed system can reduce the computation time by an approximately linear factor determined by the number of available compute clients. Compared to other related work [17][18] that focus on multiple sequence alignment and DNA sequence similarity search, we obtain similar improvement in terms of job time reduction. In addition, the scalability of the BOINC platform will allow us to maintain computational efficiency even as the size of NGS data continues to grow.…”