1. Brief BLAST revisited

We only sketch here the idea of BLAST because Wikipedia described very well. Their ideas are (in a very few words): (i) using k-mers(sometimes called k-tupples or k-letters) for the query sequence, then listing all possible matching (ii) Calculating the so-called high-scoring segment pair (HSP) (iii) Using statistical test for getting the significant matches. (and of course ones whose the small p-value (usually <0.01) will be listed as the local optimal results.

2. Discussion

- Comparison with the similar algorithm FASTA (next post)

- Various BLAST variants : NCBI maintanced a big collection, and the most recent one is Fast-BLAST (appeared in PLoS ONE last month) where they used a heuristic way to indexing the data based on the clusters and alignments from protein families).

- Any comments ?

3. BLAST vs. FASTA

BLAST and FASTA are two famous algorithms in sequence alignment. Between them, there are some similarities and differences