To elaborate, two sequences can be similar without being homologous, for instance binding sites for proteins can have selection pressures towards the same pattern. BLAST has no way to tell these similarities apart from similarities caused by shared ancestry.
Hi Biostars,
I am having a dispute with a reviewer. Is BLAST a Homology or similarity based search?
thanks
EDIT
The BLAST is performed against a custom DB containing only homologous proteins.
3 answers
...Convergent evolution
Homology is a conclusion of evolutionary relatedness. Sequence similiarity is a piece of data that would allow one to conclude homology.
BLAST might tell me that a bit of DNA is highly similar between a vector and an unfinished genome. But I'd be wrong to conclude homology from that!
statistically significant similarity implies homology
See the paper below:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3820096/
An Introduction to Sequence Similarity (“Homology”) Searching
William R. Pearson
Log in to answer this question.
The fact the your blast DB is built form homologous sequences does't change anything to answer your question. The only thing that BLAST can tell you is that your 2 (as it's pairwise alignment) sequences are similar, and the more similar they are, the less the probability you're wrong by assuming it's because those are homologs.