Profile Hidden Markov Models for Sequence Families: Identification and Alignment Strategies

1. Ch5 Profile HMMs for sequence families To identify whether a sequence belongs to a family and align it to the other members. Pairwise searching with any of the members may not find sequences distantly related to the ones you have already. An alternative approach is to use statistical features of the whole set of sequences in the search. Similarly, accurate alignment can be improved by concentrating on features that are conserved in the whole family. 1

3. 5.1 Ungapped score matrices Start by considering models for the ungapped regions.

4. 5.2 Adding insert and delete states to obtain profile HMMs

6. 5.3 Deriving profile HMMs from multiple alignments • Non-probabilistic profiles

7. Basic profile HMM parameterization The parameters to control the shape of the distribution are the values of the probabilities and the length of the model. (1) choice of length: which columns to assign to match states, and which to assign to insert states. A simple rule: columns that are more than half gap characters should be modeled by inserts. (2) probabilities: may have to add pseudo counts

9. 5.4 Searching with profile HMMs

11. Alternatives to log-odds scoring

13. 5.5 Profile HMM for non-global alignment

15. 5.7 Optimal model construction MAP match-insert assignment

17. 5.8 Weighting training sequences • Simple