We can then recover the optimal alignment by tracing back from the cell in the bottom right corner (which contains the score of aligning one complete sequence with the other) by following the pointers...We can then recover the optimal alignment by tracing back from the cell in the bottom right corner (which contains the score of aligning one complete sequence with the other) by following the pointers reflecting locally optimal choices, and then constructing the alignment corresponding to an optimal path followed in the matrix.