Project 2: Pattern Matching in Compressed DNA Sequence

Space efficient storage of large genome sequences requires good compression techniques. However, if these sequences need to be decompressed, before any processing can be done over them, the advantage of compression is lost. New techniques are required to extend the traditional pattern matching algorithms to work directly on the compressed sequence. This saves space in memory, requires less disk access and results in high speed up. In this project we will explore one such pattern matching algorithm on compressed DNA sequence, known as Derivative Boyer-Moore algorithm [2]. We will compare its running time with the traditional exact string matching algorithm, the Boyer-Moore algorithm[6], the fastest known exact string matching algorithm AGREP [3] and LZgrep, which is another algorithm that searches directly on the compressed sequence. 1 Project Description Perhaps one of the most recurrent subproblems appearing in almost every applications of computer science is the need to find the occurrences of a pattern string inside a large text. The problem is especially important in computational biology, where large sized DNA sequences are searched for finding matching patterns. Each DNA sequence contains only four alphabets A,C,T,G, but they are generally very large in size and contains vast amount of information. The human genome for instance contains three billions characters over twenty-three pairs of chromosomes. Pattern matching over such long DNA sequences requires algorithms which can handle the sequences efficiently and have very fast time complexity for search operation. In order to save storage space, it is natural to store the DNA sequences in a compressed form in a secondary storage. However decompressing them in the main memory before searching, may result in memory overflow, multiple disk access and slower running time. To overcome these adverse effects, recently there is a surge of interest in designing pattern matching algorithms which look for exact occurrences of a pattern in a compressed DNA sequence without first decompressing it. The technique allows reduction