Question : The most important aspect is the obtained result. In other words, did the rational drug design work, by indeed delivering active compounds. What is your timeline for synthesis and testing in primary hepatocyte cultures. Does Genoscience have access to in vitro protease inhibition testing ? This would be a logical first step before going to hepatocyte cultures. Answer : Time line: Modelling = 3 to 6 months Synthesis = 3 to 6 months testing in hepatocyte culture = 15 days inhibition testing = not at this time , in progress in our strategy Question : 2. In the field of modelling and computational chemistry a lot of interesting investigations have been performed. More in particular on solvatation and docking procedures. This often resulted in the production of good software. However, most of the software is not available to users in a simple way. Answer : We completely agree with your comments about these points: Solvation and docking procedures are one the specificities and concurrencial advantages of GenMol Question : 3. The GenMol modelling looks very promising according to the notes on the method. Some of the important characteristics are the flexible nature for both the ligand and the target. The actual incorporation of the solvent is also an important key topic. The check of the pdb.files, for errors, with respect to modelling is also a benefit. The study of mutant targets is also a very good approach. With the use of a universal force field a number of classic problems will not be encountered. The main conclusion is that the GenMol modelling method looks good since it contains most of the important aspects for drug design. Answer : Thanks for your appreciation Question : 4. The solvent environment is different from that in the X-ray experiment. How do you account for this problem? Answer : May be I misunderstood the question, calculations were performed with X-ray water molecules, and in different models with different number of experiment water molecules, for energy association calculations. As I observed only translations on the result values, I only kept the structural water molecule. Question : 5. Comparison of the GenMol results with other modelling techniques shows that the program is quicker and sometimes better than e.g. MSI, but the examples are limited. Do they have some other data which confirm this conclusion? Answer : GenMol goes faster than other softwares, and leads to more accurate geometries for the following reasons: - In the first steps, it uses pseudo-atoms to cross energy barriers, It goes 4 times faster than with all the atoms - In the other steps it uses local parameters, depending of the atomic and geometric neighbourhood (3000 situations are treated), that puts fastly the molecule in the probable geometry. Question : 6. Is there a particular criterium to select the tentativeactive compounds? Answer : Our criterium is the difference between the association energy and the hydation energy, the hydration energy is also used to have an idea on the possibility for the drug to reach its potential target, the molecule must be water soluble and enough lipophilic to cross the membranes. ( We report the hydration energy to the atom number, and then compar the values to those of known drugs). Question : 7. The energy difference between solvation and target interaction can be considered as a driving force. The calculated difference looks rather high. Is this checked with known examples? Answer : For HIV anti-proteases, as the protease active site is burried, the diffrence between the association energy and the hydration energy can be used as the parameter of the driving force to put the drug in the protein active site. For anti-NS3, the active site is located on the surface, it is more complicated to use this parameter, because one part of the inhibitor surface stays in contact with water. However it gives an idea on the sticking force, the true value of this difference can be obtained, in that Way: [(hydration Energy of the complex protease inhibitor + association energy of the complex protease inhibitor) - ( hydration energy of protease + hydration energy of the inhibitor)] Question : 8. With respect to the method some simple questions can be asked. One example is the dielectric constant and its function to the calculated energies, in solvent and target. Answer : The dielectric constant is distance dependant, for short contacts the correcting factor is .5. I Wrote this part of the program 20 years ago, I must check for the cutoff value. Question : 9. The cited applications all relate to modelling of enzymes and competitive inhibitors that bind in the active site and ligand-receptor docking. Does the program allow drug design based on non-active site binding inhibitors which may influence active site configurations (such asnnrti's)? 10. Would it be possible to model compounds that target the Zinc-binding site? Answer : 9- 10 The program is able to model any molecular system, the only limitation concerns the atom natures (96 atom types: all the biologic systems), Zn++ is a cation treated by the program. See attached files. Question : 11. On p19-20 and Figure 7 of "GenMol et la modélisation moléculaire" shows the small differences in structures as calculated from GenMol and as derived from the crystal structure. How do you account for the Tm of the lipids at body temperature (37°C) and how could this affect the structures of the neighbouring parts of the molecule? Answer : Tm of lipids, the program was designed to find minima of strain energies, It only uses molecular dynamics and thermalisation to get out of wrong mimima of energy and comes back to room temperature, it is not designed to predict the behaviour of molecular systems versus the temperature. Question : 12. Page 30: A number of side chains of the inhibitor are frozen onto the protease structure and modelling is then continued with mutant protease molecules. The assumptions that 1. the inhibitor binds through the same interaction sites in wt and mutant molecules and 2. the inhibitor is still active in the mutant, could be wrong. How could this impact the subsequent calculations and how do you account for this potential problem? Answer : The difference appears only in energy calculations, the lateral chains and the inhibitor are slighly modified, the energy associations between wild type and muted protease with inhibitors are very small,(which is confirmed by a thermodynamic measure on Complexes with HIV protease Protein Science 2000, 9, 1801-1809) but sufficient to distinguish activity and resistance. Question : 13. Also in the context of a highly variable virus such as HCV, it would be necessary to model a given inhibitor to numerous protease structures from the many different sequences available for HCV. As many as possible subtype 1b sequences should be tested and subsequently the selected compounds should be tested for inhibition of subtype 1a, 2b, 3a, and 4 enzymes. Do you already dispose of such data? Answer : The protease chosen was 1DY9 corresponding to the best resolution (2.1A). I remind that GenMol is automatised to generate mutants, as we do for HIV protease.