1 / 15

Class 8: Pair HMMs

Class 8: Pair HMMs. FSA HHMs: Why?. Advantages: Obtain reliability of alignment Explore alternative (sub-optimal) alignments Score similarity of sequences independent of any specific alignment. ε. W s. B q si. B (+1,+0). s ( s i ,t j ). 1 - ε. s ( s i ,t j ). δ. 1 -2 δ.

carrie
Download Presentation

Class 8: Pair HMMs

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Class 8:Pair HMMs .

  2. FSA HHMs: Why? Advantages: • Obtain reliability of alignment • Explore alternative (sub-optimal) alignments • Score similarity of sequences independent of any specific alignment

  3. ε Ws B qsi B (+1,+0) s(si,tj) 1-ε s(si,tj) δ 1-2δ Wg+Ws A psitj A (+1,+1) δ Wg+Ws C qtj 1-ε C (+0,+1) s(si,tj) ε Ws FSA  HHMs 

  4. Affine gap alignment: the full probabilistic model δ ε B qsi δ 1-2δ-τ τ 1-ε-τ Begin A psitj τ End 1-ε-τ τ 1-2δ-τ δ C qtj ε δ τ

  5. Ws B (+1,+0) s(si,tj) s(si,tj) Wg+Ws A (+1,+1) Wg+Ws C (+0,+1) s(si,tj) Ws Affine Weight Model – DP

  6. δ ε B qsi 1-2δ-τ δ τ 1-ε-τ A psitj Begin End τ 1-ε-τ τ 1-2δ-τ δ C qtj ε δ τ Viterbi in Pair-HMM • Finding the most probable sequence of hidden states is exactly the global sequence alignment

  7. δ ε B qsi 1-2δ-τ δ τ 1-ε-τ A psitj Begin End τ 1-ε-τ τ 1-2δ-τ δ C qtj ε δ τ Viterbi in Pair-HMM Initial condition: Optimal alignment:

  8. Pair-HMM for random model 1-η η Begin End η 1-η η 1-η s qsi t qtj η 1-η

  9. 1-η 1-η δ Rs1 qsi Rs2 qsi ε B qsi 1-η 1-2δ-τ δ 1-η η η τ η η 1-ε-τ η η A psitj τ 1-η 1-ε-τ 1-η τ 1-2δ-τ δ Rt1 qtj η C qtj Rt2 qtj η ε δ 1-η 1-η τ Pair-HMM for local alignment Begin End

  10. The full probability: P(s,t) Use the “forward” algorithm: The posterior probability:

  11. Suboptimal alignments Suboptimalalignments: alignments with nearly the same score as the best alignment • Only slightly different from the optimal alignment • Substantially or completely different

  12. Probabilistic sampling From the forward algorithm: Choose the next step to be:

  13. Probabilistic sampling – example

  14. Distinct suboptimal alignments Waterman and Eggert [1987]

  15. The posterior probability

More Related