1. Protein Structure Visualisation
PyMOL

2. What is PyMOL? Open-source molecular viewing program

3. PyMOL Representations Surfaces Ray-tracing (rendering)
Lines, sticks, ribbon, spheres, cartoon(s)
Surfaces
Transparency, quality
Ray-tracing (rendering)
Modes

4. Selections & Objects Every molecule (pdb file) is an object.
Selections refer to objects
Make smaller or composite objects
Changes in representation can affect objects or selections.

5. Links PDB (protein structure database) PyMOL home: PyMOL manual:
PyMOL home:
PyMOL manual:
PyMOL Wiki:
PyMOL settings (documented):

6. Aesthetics of Molecular Images
Personal taste
General guidelines:
Focus on relevant parts
Strip away unnecessary information
Find good viewing angle!
Choice of representation
No need for excessive graphics
Good figures are (almost) self-explanatory.

7. Molecular Images
Molecular graphics is an example of selective reduction of complexity.

8. Molecular Images
Molecular graphics is an example of selective reduction of complexity.

9. Molecular Images
Molecular graphics is an example of selective reduction of complexity.

10. Molecular Images
Molecular graphics is an example of selective reduction of complexity.

11. Molecular Images
Molecular graphics is an example of selective reduction of complexity.

12. Molecular Images
Molecular graphics is an example of selective reduction of complexity.

13. Molecular Images
Molecular graphics is an example of selective reduction of complexity.

14. Molecular Images
Molecular graphics is an example of selective reduction of complexity.

15. Molecular Images
Molecular graphics is an example of selective reduction of complexity.

16. Figures in PyMOL

17. Representations
ray_trace_mode 0

18. Representations
ray_trace_mode 1

19. Representations
ray_trace_mode 2

20. Representations
ray_trace_mode 3

21. Representations

22. Representations

23. Representations

24. Homology Modelling

25. No Structure = No Go? Do homology modelling. Validate the model.
Use known protein structures to make reliable models.
Validate the model.
Adjust expectations and use accordingly!

26. Basic concepts - Evolution preserves important functions
- Function is performed by specific structure
Most of the sequences don't fold

27. Basic concepts
Homologous proteins

28. Basic concepts Homologous proteins
derive from the same ancestral protein

29. Basic concepts Homologous proteins
derive from the same ancestral protein
Orthology occurs
Paralogy occurs

30. Basic concepts Homologous proteins
derive from the same ancestral protein
Orthology occurs with Speciation
Structure and function very conserved
Paralogy occurs with Duplication
Structure conserved

31. Homology modeling Identify a homolog with solved structure
(structural template)
using sequence similarity
to copy the conserved regions

32. Homology modeling Identify a homolog with solved structure
(structural template)
using sequence similarity
to copy the conserved regions
Structure and function are more conserved than sequence

33. How Often Can We Do It? Currently ~100,000 structures in the PDB
Approx. ~12000 with sequence ident. <30 % and resolution <3.0 Å.
25% of all sequences can be modelled.
50% can be assigned to a fold class.

34. How Well Can We Do It?
Sali, A. & Kuriyan, J. Trends Biochem. Sci. 22, M20–M24 (1999) 

35. How Is It Done? Identify template(s) Improve alignment
Initial alignment
Improve alignment
Backbone generation
Loop modelling
Side chains
Refinement
Validation 

36. Template Identification
Search with sequence
Blast (>60%)
Psi-Blast (>30%)
profile-profile alignment (>15%)
Fold recognition methods (>10%)
Use biological information
Functional annotation in databases
Active site/motifs

37. alignment pipelines Hhpred (profile-profile comparison)
3 psiblast/hhblits cycles (improve alignment locally with mac)
- find homologs with 2 cycles of cs-blast realign with muscle/mafft check ins/dels use the alignment with HHpred

38. Alignment

39.
PHE ASP ILE CYS ARG LEU PRO GLY SER ALA GLU ALA VAL CYS
PHE ASN VAL CYS ARG THR PRO GLU ALA ILE CYS
PHE ASN VAL CYS ARG THR PRO GLU ALA ILE CYS F D I C R L P G S A E V 6 -2 -3 2 -1 N 1 5 8 T

40.
PHE ASP ILE CYS ARG LEU PRO GLY SER ALA GLU ALA VAL CYS
PHE ASN VAL CYS ARG THR PRO GLU ALA ILE CYS
PHE ASN VAL CYS ARG THR PRO GLU ALA ILE CYS F D I C R L P G S A E V 6 -2 -3 2 -1 N 1 5 8 T

41. Improving the Alignment
PHE ASP ILE CYS ARG LEU PRO GLY SER ALA GLU ALA VAL CYS
PHE ASN VAL CYS ARG THR PRO GLU ALA ILE CYS
PHE ASN VAL CYS ARG THR PRO GLU ALA ILE CYS
From ”Professional Gambling” by Gert Vriend

42. Thumb rules no deletion with terminals far apart in space
no insertion in the core
check for conserved residues, especially G, P, C
check for motif conservation
be careful with secondary structure prediction
identify disordered/accessory regions

43. Template Quality Selecting the best template is crucial!
The best template may not be the one with the highest % id (best p-value…)
Template 1: 93% id, 3.5 Å resolution 
Template 2: 90% id, 1.5 Å resolution 

44. Backbone Generation
Generate the backbone coordinates from the template for the aligned regions.
Several programs can do this, most of the groups at CASP6 use Modeller:

45. Side Chains
Side chain rotamers are dependent on backbone conformation.
Most successful method in CASP6 was SCWRL by Dunbrack et al.:
Graph-theory knowledge based method to solve the combinatorial problem of side chain modelling.

46. Side Chains Prediction accuracy: High for buried residues
Low for surface residues
Experimental reasons: side chains at the surface are more flexible.
Theoretical reasons: much easier to handle hydrophobic packing in the core than the electrostatic interactions including H-bonds to waters.
Myoglobin

47. Side Chains
If the seq. id is high, the networks of side chain contacts may be conserved.
Keeping the side chain rotamers from the template may be better than predicting new ones.

48. Error Recovery
Errors in the model can NOT be recovered at a later step
The alignment can not make up for a bad choice of template.
Loop modeling can not make up for a poor alignment.
The step where the errors were introduced should be redone.

49. Model refinement Close loops with Rosetta, modeller, FREAD
SCWRL4 for side chains
Check for error/clashes before refining
don't over-refine (KobaMin)

50. Validation Most programs will get the bond lengths and angles right.
Model Rama. plot ~ template Rama. plot.
select a high quality template!
Inside/outside distributions of polar and apolar residues.

51. Model Validation – ProQ
ProQ is a neural network-based predictor
Structural features  quality of a protein model.
ProQ is optimized to find
correct models…
…NOT (necessarily) native structures.
Two quality measures:
MaxSub & LGscore
Arne Elofssons group:

52. Structure Validation Program Suites
ProCheck
WhatIf server