Protein Conformation Prediction (Part II)

1. Doug Raiford Lesson 18 Protein Conformation Prediction (Part II)

2. Review: two folding models • Framework model • Secondary structure first • Assemble secondary structure segments • Hydrophobic collapse • Molten: compact but denatured • Formation of secondary structure after: settles in • van der Waals forces and hydrogen bonds require close proximity Protein Conformation Prediction (Part II)

3. Two approaches • De novo (or ab initio) • From the beginning or from first principles • Comparative/Homology Based • Sequence similarity Protein Conformation Prediction (Part II)

4. Homology based • Find a similar protein of known structure • Structure should be similar Protein Conformation Prediction (Part II)

5. How • Know the phi and psi angles of the similar protein • Can apply those same angles • Known as threading Protein Conformation Prediction (Part II)

6. Threading issues • What are chances that lengths will the same • Where put longer portions • Where put gaps Once again, MSAs Protein Conformation Prediction (Part II)

7. Popular homology based approach • 3D PSSM (Threading Server) • Remember? • Position specific similarity matrix • Profiles • 3D PSSM performs MSA’s but augments with additional 3D alignments • Aligning “known 3D conformations” in three dimensions Protein Conformation Prediction (Part II)

8. De Novo Approaches • Molecular dynamics • Summation of all forces exerted at all locations simultaneously • Computationally intensive • Do not fully understand such forces as hydro-phobic avoidance of solvent Protein Conformation Prediction (Part II)

9. Middle ground • Secondary structure prediction • Accuracy mid to upper 70’s • Work the loops to fold secondary structures into energetically optimal conformation Protein Conformation Prediction (Part II)

10. One approach… • See how often aa’s show up at specific positions in secondary structure • Chou-Fasman • Empirical parameters for , , and  -turns • P(,aa)=f(aa)/ave() • P(,aa)=f(aa)/ave() • P(-turn,aa)=f(aa)/ave(-turn) • Name P(a) P(b) P(turn) • Alanine 142 83 66 • Arginine 98 93 95 • Aspartic Acid 101 54 146 • Valine 106 170 50 Protein Conformation Prediction (Part II)

11. Algorithm • ID regions where 4 out of 6 (3 of 5 for ) contiguous residues have P(a-helix) > 100 • Extend the helix in both directions until a set of four contiguous residues that have an average P(a-helix) < 100 is reached. 1 MAKYNEKKEK KRIAKERIDI LFSLAERVFP YSPELAKRYV ELALLVQQKA HHHHH HHHHHHHHHH H HHHHHHHH HHHHHHHHHH 51 KVKIPRKWKR RYCKKCHAFL VPGINARVRL RQKRMPHIVV KCLECGHIMR T SSTTTT SB TTT B BTTTEEEEE E SSS EEEE EETTTTEEEE 101 YPYIKEIKKR RKEKMEYGGL VPR EE Protein Conformation Prediction (Part II)

12. For turns • Chou and Fasman also determined turn frequencies • Most hairpins are three in length • When p(-turn) = f(j)f(j+1)f(j+2)f(j+3) is greater than P() or P() • Name P(a) P(b) P(turn) f(i) f(i+1) f(i+2) f(i+3) • Alanine 142 83 66 0.06 0.076 0.035 0.058 • Arginine 98 93 95 0.070 0.106 0.099 0.085 • Aspartic Acid 101 54 146 0.147 0.110 0.179 0.081 • Valine 106 170 50 0.062 0.048 0.028 0.053 Protein Conformation Prediction (Part II)

13. Patterns in usage • Patterns can be used to augment these statistical approaches • In some cases, one side of helices like water • Every 4th aa hydrophilic • Helps ID helix • Helps ID that solvent exposed • Other patterns: coiled coils Protein Conformation Prediction (Part II)

14. Sounds like… • Does this sound familiar? • Probability of a sequence of occurrences? Hairpin position 1 Hairpin position 2 Hairpin position 3 Protein Conformation Prediction (Part II)

15. HMMSTR • Hidden Markov Model • Hidden states helix, beta sheet, turn Protein Conformation Prediction (Part II)

16. Motifs • Proteins organized into • Domains • Domains composed of motifs • PFAM • Database of protein families • Hidden Markov Models Protein Conformation Prediction (Part II)

17. HMMR and PFAM Protein Conformation Prediction (Part II)

18. CASP • Critical Assessment of techniques for protein Structure Prediction • Biannual conference: contest • Secret newly experimentally determined structures CASP1 (1994) | CASP2 (1996) | CASP3 (1998) | CASP4 (2000) | CASP5 (2002) | CASP6 (2004) | CASP7 (2006) | CASP8 (2008) | CASP9 (2010)| CASP10(2012) Protein Conformation Prediction (Part II)

19. CASP evaluation • Root mean square (RMS) for angles • No intermol contacts • Secondary structure • Surface • Buried Protein Conformation Prediction (Part II)

20. Approaches • Have seen comparative homology based • HMM based rely on multiple sequence alignments: homology • Now turn to De novo • Split into two: Ab initio and knowledge based Protein Conformation Prediction (Part II)

21. ROSETTA (CASP3 Winner) De Novo: Knowledge based • Build a list of possible conformations (25) for each segment (length 9) • Predicted secondary structure • Database of structures • Randomly draw from this list, apply ψ and φ, and score conformation • Monte Carlo simulated annealing procedure Protein Conformation Prediction (Part II)

22. ROSETTA (CASP3 Winner) • Scoring global conformation • hydrophobic burial • Electrostatics • Disulfide bonding • Main chain hydrogen bonding • Strand pairing • Sheet formation • Helix-strand interactions • Excluded volume Protein Conformation Prediction (Part II)

23. Protein Conformation Prediction (Part II)

24. Protein Conformation Prediction (Part II)