1. Bio/Chem-informatics
© José R. Valverde, 2014
CC-BY-NC-SA

2. From sequence to atoms Cheminformatics

3. Index Goals Obtaining protein structures Obtaining protein sequences
Comparing structures and sequences
Obtaining ligand structures
Limits

4. Goal Learn as much as you can about your protein
Identify relevant properties
Function
Active site(s)
Modifications
Conserved features
Relevant amino acids
Cheminformatics: the application of informatic methods to solve chemical problems

5. Read Bibliography http://www.ncbi.nlm.nih.gov/pubmed
Should be the initial step in all cases
Should have been already done
Likely to be neglected
It is funnier to play from the start
Guides all subsequent analysis and experiment
Allows taking a decision
It IS worth the trouble!

6. Sequence analysis
Compare sequences and look for similarities and differences
Match to experimental observation

7. Predict, predict, predict...
Secondary Structure
Properties (ProSite, PFAM, InterPro...)

8. ProSiteDoc {PS00433; PHOSPHOFRUCTOKINASE} {BEGIN}
*********************************
* Phosphofructokinase signature *
Phosphofructokinase (EC ) (PFK) [1,2] is a key regulatory enzyme in
the glycolytic pathway. It catalyzes the phosphorylation by ATP of fructose
6-phosphate to fructose 1,6-bisphosphate. In bacteria PFK is a tetramer of
identical 36 Kd subunits. In mammals it is a tetramer of 80 Kd subunits. Each
80 Kd subunit consist of two homologous domains which are highly related to
the bacterial 36 Kd subunits. In Human there are three, tissue-specific, types
of PFK isozymes: PFKM (muscle), PFKL (liver), and PFKP (platelet). In yeast
PFK is an octamer composed of four 100 Kd alpha chains (gene PFK1) and four
100 Kd beta chains (gene PFK2); like the mammalian 80 Kd subunits, the yeast
100 Kd subunits are composed of two homologous domains.
As a signature pattern for PFK we selected a region that contains three basic
residues involved in fructose-6-phosphate binding.
-Consensus pattern: [RK]-x(4)-G-H-x-Q-[QR]-G-G-x(5)-D-R
[The R/K, the H and the Q/R are involved in fructose-6-P
binding]
-Sequences known to belong to this class detected by the pattern: ALL.
-Other sequence(s) detected in Swiss-Prot: NONE.
-Note: Escherichia coli has two phosphofructokinase isozymes which are encoded
by genes pfkA (major) and pfkB (minor). The pfkB isozyme is not evolutionary
related to other prokaryotic or eukaryotic PFK's (see <PDOC00504>).

9. InterPro Database of protein families, domains and functional sies
Integrates other databases: PROSITE, PRINTS, ProDom, SMART, TIGRFAMs, PIRSF, SUPERFAMILY, PANTHER, GENE3D...

10. InterProScan

11. PredictProtein
Automatic prediction of structural and functional properties of proteins
Runs a test battery
And gives a detailed
report

12. Look for known structure

13. Search for homologs Search in structural databases PDB/RCSB
Search in Sequence databases
Blast against SwissProt
Blast against EMBL/GenBank/DDBJ

14. Blast vs. PDB (EBI)
Search for sequence-related structures

15. NCBI BlastPDB
Search for structures of sequence-related structures

16. ModBase
Search for possible 3-D models of the protein

17. Nature's SBKB
Search for models from a number of servers

18. Alignment of mt ATP6
Spot a few, well-preserved, amino acids with a major role.

19. Multiple Alignment Problems
Homologue proteins
Risk: Too high conservation
Same family
Risk: Too little conservation

20. Analyze coevolution Co-evolving amino acids highlight interactions
See review at CNB

21. Structural matching Protein Function Prediction Server
Uses structural data from known files to make predictions
Catalytic Site Atlas
Uses structural
models of active
sites

22. Compare, compare, compare...
The answer may already be there
If not, similarities and differences allow you to scan genomes for useful targets, and proteins for target sites.
There are many tools.
There are “supertools” combining many tools
e.g. STING Millenium
Information is often cheaper than calculation

23. Limits Still reduced knowledge of 3-D structures
Prediction accuracy needs to be asserted
Check the database metadata
Available models may be outdated or incorrect
Too high or too low conservation preclude specific assignment
New, unknown proteins and functions are possible

24. But, wait! There is more... much more!
Image by geralt. CC0.