Drug discovery 3&4 Flashcards
What is CASP?
Blind trial for testing structure prediction algorithms of determined structures before publication
How often is CASP performed?
Every two years
In energy minimisation is water a factor?
No
In Molecular dynamics is water a factor?
Yes
What are the limitations with energy minimisation?
Energy landscape not known hence difficult to identify lowest energy fold.
Also even if lowest energy minima found the protein may not exist in this structure
The accuracy of a 3 state prediction is known as what?
Q3
If secondary structures are assigned randomly what would the value Q3 to be?
33%
What was the function of Chou-Fausman?
To predict secondary structure
What is Chou-Fausman based on?
propensity of amino acid to exist in certain secondary structures. Each amino acid is given a score based on its propensity to exist in secondary structure
What is an average propensity of amino acids in Chou-Fausman?
1
1 < likely
1 > unlikely
What are the rules of Chou-Fausman?
A run out of 4 out of 6 residues favour helix then there will be a helix predicted.
Extend helix until a proline is reached or a run or 4 with propensity lower than 1
What is the principle of stereochemical method of secondary structure prediction developed by Lim?
Recognition of patterns of hydrophobic/hydrohpillic residues that favour secondary structures
Why is the stereochemical algorithm developed by lim empirical?
It is enhanced by inspection
What was the Q3 of the stereochemical algorithm developed by lim empirical?
60% - poor
What is an artificial neural networks?
Machine learning algorithm for assigning secondary structure after a training stage.
What was the Q3 of artifical neural network?
62%
How does MSA help with secondary structure prediction?
Highly conserved regions identifies important regions.
e.g. conserved sequence of aa of high alpha helix propensity is more likely to be an alpha helix
What was the Q3 when MSA is used?
77%
What is homology modelling?
Prediction a estruture based on a known structure of a similar protein
What is the first step of homology modelling?
Blast/PSI-blast
What follows the blast/psi-blast search in homology modelling?
Alignment of sequences of unknown with known (create MSA) - can be helpful to create alignment based on secondary structure prediction
What do you do if your MSA produced in homology modelling has gaps?
- Search for equivalent loops in PDB to fit the structure
- Energy minimisation of the loop
What type of loop is the best to predict?
Short loops (<6 residues), well predicted by all methods
After you substitute the sequence from known structure with the unknown structure and fit loops in what is the next step of homology modelling?
Pack side chains - given the backbone predict psi, phi, omega, chi
Only allowed angles need to be sampled
