Sequence Comparison Algorithms in Genome Databases

We surveyed several important formulations of the sequence database search problem (i.e. local sequence alignment) as well as their solutions. These include the Maximal Segment Pair (MSP) formulation, which has a trivial dynamic programming solution and serve as the basis for popular heuristic solutions such as BLAST 2] and FASTA 18]. The local alignment with gaps formulation as well as a restricted version of the problem is also studied in detailed. Several recent linear and sub-linear time algorithms are studied and compared. A theory for alignment scores based on the MSP formulation is examined. The theory gives a framework to choose the substitution matrix under a particular context. Under the same framework, the statistical signiicance of alignment scores can be assess. BLAST is a fast, and popular heuristic solution that combines ideas from formal approaches. We outline the main algorithm as well as the reasoning behind the choice of the parameters through formal theoretical analysis and empirical evidence.