hmms (``HMM search'') calculates the probability of a global alignment of the target sequence to the model; i.e. the whole model must match the whole sequence, as in Needleman-Wunsch style alignment [11]. This is used when one has a model of a whole protein, such as the globins, and expects to find only complete end to end matches in the database.
The command line and options are identical to hmmsw and hmmfs, except that the -c option for searching a complementary strand isn't available. The output is simply the score of the match and the name and description of the matched target sequence.
> hmms globin2.hmm bashford.slx
hmms is usually only useful if you're checking the scores of a model against a training database of sequences, or if you're really sure that the model always matches across its full length. For almost all protein database searches, hmmsw is preferable.