HPC Solution for the Drug Repositioning Problem

Transcription

1 HPC Solution for the Drug Repositioning Problem Andrea R. Beccari Drug Discovery Platform Manager

2 Summary R&D Challenge in Pharmaceutical industry Computational Chemistry in Pharma R&D LiGen HPC Application 2

3 R&D in Pharmaceutical industry Productivity Nature Reviews Drug Discovery 11, 6-8 (January 2012) 3

4 R&D in Pharmaceutical industry Efficiency 4

5 R&D in Pharmaceutical industry Probability of Transition: (pts) 4 product enter the market out of 100 R&D discovery programs Nature Reviews Drug Discovery 9, (March 2010) 5

6 R&D in Pharmaceutical Industry Challenges Reduce failure rate 24 programs for 1 market drug Reduce time From 10 to 13 years for 1 market drug Reduce costs From 0.8 to 2 billions for 1 market drug 6

7 R&D in Pharmaceutical industry Drug Repositioning Definition Use a market drug for another indication Pro Reduced time to market and failure rate Con The molecule could not be modified 7

8 Computational Chemistry in Pharma R&D Computational chemistry supports all the phases of drug discovery process: HPC allows to Use more complex theoretical models Dramatically reduce calculation time Simulate huge systems 8

9 Computational Chemistry in Pharma R&D Need for better models and tools 4 th Call of IMI: Technology and Molecular Disease Understanding Understanding and optimizing binding kinetics in drug discovery Rationale: Kinetic of binding of candidate drugs strongly correlates with efficacy in human Budget: > 100 Mio euro 9

10 LiGen Why Develop a new tool? Commercial and open source programs do not full fill all of our requirements Use proprietary technologies is a competitive advantages Vs competitors None of the other tools are designed to run on real HPC architectures 10

11 LiGen The Team Dompé [Since 2005] Andrea R. Beccari Simone Lorenzi (Docking Specialist) CINECA [Since 2005] Carlo Cavazzoni (Super Computing Applications) Andrew Emerson (High Performance Systems) University of Parma [Since 2010] Gabriele Costantino (Full professor) Claudia Beato (Design of Experiment) 11

12 LiGen What is Molecular Docking In the field of molecular modeling, docking is a method which predicts the preferred orientation of one molecule to a second when bound to each other to form a stable complex. Knowledge of the preferred orientation in turn may be used to predict the strength of association or binding affinity between two molecules using for example scoring functions. 12

13 LiGen Map The Cavity of the Protein 13

14 Spring Docking Flexible Pharmacophoric Docking 14

15 LiGen Docking Result 15

16 Design of Experiment exp01 exp04 exp07 exp10 exp13 exp16 exp19 exp22 exp25 exp28 exp31 exp34 exp37 exp40 exp43 exp46 exp49 exp52 exp55 exp58 exp61 exp64 exp67 exp70 exp73 exp76 exp79 exp82 16

17 Benchmark % RMSD < 3 A Average pocket time (s) Average dock time(s) Average Total time (s) Glide XP 72,4% 161,42 173,92 335,34 Ligen 75.5% 9,30 41,0 50,30 Autodock 85,3 % 9,83 397,08 406,91 Dataset: Redocking of 171 complexes (PDBbind core set and MW < 500) 17

18 LiGen Intel Beta testing Knights Ferry Beta Application Decision: Approved! LiGen modules have been designed to be at the same time flexible and highly efficient from a computational point of view (use of C++ and STL at high level and proprietary C/C++ algorithms at low level), with particular attention to modern and future Many Integrated Core (MIC) architecture. In agreement with Intel a porting to experimental Knights Ferry processor is ongoing, which represents an extremely promising platform for HTVS application 18

19 LiGen The Drug Repositioning Problem The Problem Find a market drug that effectively bind a biological target. Costs for performing real experiments 1 market drug Vs 200 Biological Targets: 40K euro 8K market drugs Vs 200 Biological Targets: 320M euro Number of experiments: 1.6M Success rate : 0.5% 19

20 LiGen HPC Solution for Drug Repositioning Problem Cost for performing docking simulations 16K Near to Market 8K Market Drugs 30K Targets 3M hours 3M hours = 300K euro Number of simulation = 720M Success rate = 0.01% 20

21 Preliminary Results During the Validation We have selected 32 market drugs coming from molecular docking simulations and we have experimentally tested against TRPM8 that is a relevant target for prostatic cancer. 1 market drugs shows, a never reported before, relevant biological activity on this target. Patent filing is on going. Title 21

22 Thank You! 22