[DIYbio] Re: Chemical Structure Generation for Drug Simulation

Hi,

First off, a word of caution - the problem you're looking at here is worthy of multiple PhD theses. That being said, ambition is good, right? :)

1. When you say you want to generate all possible combinations of chemicals, you'll need to be more specific with your question. Do you have a specific small molecule drug compound in mind, and you want to generate different conformations of that compound? Or do you want to traverse through the space of all possible small molecule drugs that could exist? If the latter, I would suggest that this is impossible, because the space approaches infinity combinatorically rather quickly as you increase the number of allowed atoms.

A good algorithm that searched this space intelligently, however, and could provide suggestions for new, unknown small molecule chemicals, could be quite beneficial.

2. In order to predict activity with MD, you need to have prior experimental knowledge of what protein states are active and which are inactive (at least as far as any studies I've seen). This is because the activity of the protein is not something that is immediately obvious outside of its context. This is especially true of proteins that interact with other proteins or have activity that occurs over the time scale of milliseconds or longer.

Instead, researchers usually target a known binding site, and try to measure binding affinity for their target compounds. This is a costly MD experiment to run, so usually it is only done on a relatively small number of compounds, following a heuristic funnel that reduces the number of targets from millions to dozens.

3. For examples of proteins and their interactions - have you tried the PDB? http://www.rcsb.org/pdb/home/home.do

Best,
Tom
PhD, Computational Biology, NYU '12

On Friday, March 13, 2015 at 1:43:56 PM UTC-4, Gokula Krishna wrote:

Hello there,

I am a python programmer who is interested in automating the drug simulation process. I taught myself about the basics of:

1. Molecular Dynamics

2. Pharmaco Kinetics

3. Pharmaco Dynamics

4. PyMOL

I am planning to create a program that generates all possible combinations of chemicals for the small molecule drug for a given protein. I could not find any source related to the following:

1. How to generate all possible combination of chemicals of drug for the protein? And how the generated chemicals should be represented for the protein-chemical Interaction (chemical structure like CH4 or other format)? 

2. How to determine whether the given chemical is an agonist or antagonist computationally? And how to determine whether the viral protein is deactivated for the particular chemical?

3. An example on interaction between a simple viral protein and a chemical drug which deactivates it.

Thanks,