Computational Biology Lecture 6: Affine Gap Penalty Function, Multiple Sequence Alignment

We saw earlier how we can use a concave gap penalty function γ, i.e. one that satisfies γ(x+1)−γ(x) ≤ γ(x)−γ(x−1), as a more realistic model for penalizing gaps. This was at the expense of increasing the time complexity of our dynamic programming algorithm. With a concave gap penalty function, the new DP algorithm requires O(mn + nm) time, as compared to the O(mn) time bound of previous algorithms. We will seek a solution that will bring us back to our old time bound of O(mn) while still providing a good model for gap penalty.