I'm searching a set of bacterial proteomes for a family of genes that undergo frequent HGT and domain restructuring, such that homologs in closely related species may appear quite divergent. My intuition tells me that a profile hidden Markov model approach such as HMMER3 may be better than using standard BLAST-based pipelines. I understand that BLAST is a heuristic searching tool whereas HMMER scores matches using a formal probabilistic basis, which has its advantages in terms of reproducibility, but I was wondering if anyone was able to tell me what other advantages / disadvantages may exist between the two methods? And if there are specific types of analyses where one method has clear advantages over the other? Any links to references etc. would also be much appreciated.
Thanks in advance!