Hi Rajesh , you have raised a good question. I have identified few kinase inhibitors trough pharmacophore search and molecular docking and few of them are active in in -vitro kinase assay. If your protocol is right then you will definitely get few active molecules against your target/targets. When we start with docking , we rely on docking energy as well as docked conformation. However we cannot completely beleive in docking reults because most of time we select those which favours our findings. Important point here is , how we validate our doking results . We can use different docking softwares to check that consistent results are coming or not. We can do MD simulation to check the stability of bound conformation over time. We can aslo do free energy analysis etc. I think if you use integrated approach then you wil have good finding. Designing a selective inhibitor is very tough job so first you proceed with experimental validation of few slected hits and then go for optimization.

Here is a link , you may find helpful.

http://www.ncbi.nlm.nih.gov/pubmed/15546735

HI,

Nice question and the answer is difficult. First, try to do clustering for your compounds and only synthesize those having diverse structures until you find a good hit to start your focus (Lead or hit generation) which i think is the painful part if you are trying to identify anew scaffold. As Kumar said, score your molecules with more than one scoring function to get the best out of each and do not follow docking blindly. Once you got a good hit (10-20 microMolar IC50) you can build a more focused library. For this step (Lead optimization) you may need to use a more rigorous scoring technique, such as TI or MMPSA. You need also to check stability as i feel it is sometimes even more important than the binding energy itself. For kinases in particulars, see how the conserved water molecule might behave.

One caveat here is that your ligand -as you said- should be new so you should pay attention how it might bind, may be it binds in a fashion that is different from the conventional quinazoline based ones..This is very difficult to prove especially if you are planning for a good Journal. J Med Chem, for example, asks for X-ray crystal structure as a proof.

Good Luck

Hi, EGFR TKIs are one of widely studied molecules for cancer chemotherapy. However, you have to read the literature to understand if they have been successful or had a good therapeutic index.

If your motive is to generate some data for a publication, the approach followed by you is Ok. Vikash Kumar and Marawan Ahmed have adequately cautioned you about the fickleness of docking scores and some trends in the literature. How does the docking score of erlotinib, your reference molecule, compare with the scores of your short listed chromenone library? This should help you to select the molecules for synthesis.

The kinase domain in EGFR is located on the inside of the cell membrane. Therefore, any inhibitor that you may design, which you want to work in an in vivo (cell-based assay) will have to have adequate characteristics to cross the cell membrane. Also, many EGFR TKIs can cross dock with the other GFR TKs such as VEGFR TK and PDGFR TK. So, do a quick survey of the landscape before deciding the line of your action. All the best.

Dear Marawan and Gopalakrishnan,

Thanks for your suggestions. Marawan has quoted scoring the molecules with more than one scoring function to get the best out of each and to get good hits. During docking I first refered the interactions of erlotinib in cocrystallized structure. I docked erlotinib in protein by changing various parameters, runs etc. and stopped when I got same type of interaction, H-bond length and same type of conformer as in crystal structure. Then I docked the designed ligands with this parameter constant. I almost all cases I found that designed molecules bind in quite similar manner as that of erlotinib (Viewed in Pymol). Now other than this how should I use different scoring functions I am not getting. Will I get more help at autodock forum?

Gopalkrishnan has suggested to survey, in this case is it ok if I carryout docking of some prominent molecules with all GFRs and decide the final target? I have carried out the ADMET exhaustively and I am having now the details how my molecules are better than the erlotinib. In case of assay also I am stuck in which enzyme assay should I choose. I have done studies on few cell lines and got encouraging results. And thanks for discussing ...

Hi Rajesh,

Your target should depend on which assay you have or wish to do. If you have assays and want to do all the GFR TKI experiments, you would have done a good study! If you have many ligands that are better than erlotinib in terms of docking score, select a few that you can synthesise and check their activity first.

HI Patil,

I had a very little experience with autodock but i can remember it uses different types of scoring functions and then you can use any (or all) of them to test your hypotheses. One trick i usually use sometimes is trying to identify what is the best term of the scoring function itself that can explain the trend in the data. Scoring functions are usually composed of different terms (eX: vdW, electrostatic, H-bond..etc) and the weight of each of them in the final scoring function is determined by the coefficient of that term. These coefficients are mostly fitted against a wide range of experimental binding energy data. And due to certain experimental uncertainties and the different contribution of each term across different protein targets, may be in your case these weights can have different values, so you might need to optimize these parameters using linear regression or PLS schemes. This is the spirit of Linear Interaction Energy models if you are aware of. You can also correlate your experimental data against each term of the scoring function and see which one has the highest correlation. I have used these tricks in some of my papers and they are in my page if you are interested. Usually in kinases, lipophilic interaction achieves the best correlation. You can access this using Pearson correlation in Excel.

Cheers

Marawan

Thanks Maravan, I wasn't aware of such elaboration of influence of scoring functions. Earlier my assumption was that different software uses different scoring functions (ex. Autodock uses Amber, Dock uses emperical, Vlife Uses MMFF etc.) and docking experiments with these different software would give me different results based on these scoring functions. I carried out docking in Autodock4, Vina, Vlife sciences and Dock for my molecules. I don't know much about finding weight of each scoring function and their optimization by regression analysis. You page has wealth of information on this and I am going to acquire it. For now I am going to follow Gopalakrishnan's suggestions as they are simple and practicable for me at this hour. Thanks

Thanks Gopalakrishnan

For the article