What is the computational complexity of the best known algorithm for computing the similarity between two sequences (as in DNA or Protein alignment/approximate string matching)?
The similarity is based on:
scoring the alignment using substitution scoring matrices (for either global or position-specific substitutions of 20 symbols in Protein alphabet or 4 symbols in DNA alphabet)
Is the linear time of Burrows–Wheeler transform used in Bowtie and BWA short-read aligners the actual state-of-the-art or are there sub-linear algorithms solving the same problem?
[Edit]: Thinking of applying LSH for approximate matching that will be sublinear assuming pre-processing/indexing of the reference dataset