A study on inertia weight schemes with modified particle swarm optimization algorithm for multiple sequence alignment

In this study a modified particle swarm optimization (MPSO) algorithm has been proposed for obtaining optimal multiple sequence alignment (MSA). Due to the NP-complete nature of MSA problems, it is always a challenging task to obtain a suitable approach for optimal alignment of specific family of sequences. Proposed MPSO approach takes the random sequence length (with gaps), with non-repetitive gap positions in each iteration. Proposed approach applies both integer and binary matrix for incorporating gaps in sequences so as to avoid overlapping of gaps with sequences. Seven inertia weight schemes have been tested so as to check the effect of weight on convergence speed and time. Here the weight effect has been studied for different problem sizes i.e. 5 & 10 sequences with maximum sequence length of 10 & 212 respectively. MPSO is found to be performing well than standard PSO (S-PSO) at the same parameter set. Exponentially decreasing weight scheme is found performing significantly better at time and convergence criteria than other six weight approaches.