1. PREDICCIÓN DE ESTRUCTURAS DE CRISTALES CON MOLÉCULAS FLEXIBES EN SU CELDA V. Bazterra, M. B. Ferraro, J. C. Facelli Departamento de Física Facultad de Ciencias Exactas y Naturales Universidad de Buenos Aires 2007

2.  Crystal engeneering  Pharmaceutical design  Polymorphism  Application in materials. AIM OF THE APPLICATION

3. Why GENETIC ALGORITHMS?  Useful to model atomic and molecular clusters.  Difficult crystal prediction from first principles.  Polymorphic forms in organic crystals

4. MGAC Crystal Structure Prediction Capabilities Victor E. Bazterra, Matthew Thorley, Marta B. Ferraro, and Julio C. Facelli J. Chem. Theory Comput. 2007, 3, 201-209 Search for crystal structures within any symmetry group and with an arbitrary number of molecules and molecular types per asymmetric unit. Search for crystal structures within any symmetry group and with an arbitrary number of molecules and molecular types per asymmetric unit. Search structures using either the rigid or flexible molecule models. Search structures using either the rigid or flexible molecule models. Automatically generate the molecule’s force field using existing force field libraries. Automatically generate the molecule’s force field using existing force field libraries. Increase the sampling power and the complexity of molecules amenable to CSP studies using the parallel and distributed computing capabilities of the system. Increase the sampling power and the complexity of molecules amenable to CSP studies using the parallel and distributed computing capabilities of the system. Automatically compare, sort and archive the most relevant structures in a user database. Automatically compare, sort and archive the most relevant structures in a user database.

5. GENETIC CODING Molecular center of mass  {R 1, R 2,…R n }  Its orientations  {  1,  2,…  \n } Relevant dihedral angles  {  1,  2, …..  n } Space group and lattice parameters  {a,b,c, , ,  } (Rigid bodies)

6. Genetic Algorithms Application

7. SEMI-RIGID APPROXIMATION DATA -Crystallographic group -Number of molecules in the cell PARAMETERS -Lattice angles, , ,  -Lattice axis a, b, c APTITUDE FUNCTION -AMBER FORCE FIELD -CHARMm FORCE FIELD in CHARMM code MOLECULES: -center of mass positions -relative orientations

8. SEMIRIGID APPROXIMATION Rigid bodies with flexible chains Parameters to be optimized -crystallographic cell axes and angles. -positions of the center of masss of each molecule. -Euler angles respect to the unit cell. -N dihe molecular angles. K=6+Z(6+N dihed.)

10. Interface with CHARM Module

11. Evolution of the population energy Hystogram of the evolution Population analysis

12. Comparison between crystals

13. Fragment matching

14. Local optimization using CHARMM 6, 7 with the GAFF 14 parameters. cutoff of 14 Å, and the electrostatic interactions were calculated using the Ewald technique. Local optimization using CHARMM 6, 7 with the GAFF 14 parameters. cutoff of 14 Å, and the electrostatic interactions were calculated using the Ewald technique. Atomic charges,, using the restrained electrostatic potential approach implemented on the RESP program. Gaussian03 32 package at HF/6- 31G* level. Atomic charges,, using the restrained electrostatic potential approach implemented on the RESP program. Gaussian03 32 package at HF/6- 31G* level. Restricted searches using 30 individuals in the population for up to 130 generations, for the 14 most common symmetry groups for organic molecules, P1, P-1, P21, C2, Pc, Cc, P21/c, C2/c, P212121, Pca21, Pna21, Pbcn, Pbca and Pnma. Restricted searches using 30 individuals in the population for up to 130 generations, for the 14 most common symmetry groups for organic molecules, P1, P-1, P21, C2, Pc, Cc, P21/c, C2/c, P212121, Pca21, Pna21, Pbcn, Pbca and Pnma. For each molecule we performed between 150 and 200 runs leading to at least 100 complete runs with 130 generations. For each molecule we performed between 150 and 200 runs leading to at least 100 complete runs with 130 generations. From these short lists we manually detected clearly unphysical structures and duplicated ones that were not eliminated in the previous step that were identified by comparison of their XRPD spectra From these short lists we manually detected clearly unphysical structures and duplicated ones that were not eliminated in the previous step that were identified by comparison of their XRPD spectra CSP2007 Methodology

15. Molecule I

17. Molecule III

19. Molecule XII ACRY02 Space group: Pbca

20. Molecule XII EXP. Acry02 (Pbca) Predicted Crystal_051_050 (Pbca) a 6.764 (6.970) 6.764 b 9.866 (9.752) 9.866 c 9.536 (9.514) 9.536 9090 9090 9090 RMS (Å) 0.245817 ENERGY (kJ/mol) -22.402-22.402

21. Molecule XIV P21/c

22. Molecule XIV Exp. Alphyph (P21/c) Predic. Crystal_06_006 (P21, P21/c) a 14.042 (13.060) 14.882, 14.046 b 9.613 (9.738) 9.612, 9.612 c8.264 ( 9.335) 9.551, 8.263 90 90, 90  100.9 (105.8) 156.942, 100.94 90 90, 90 RMS (Å) 0.830175 ENERGY (kJ/mol) - 19.660 -19.675

23. ?????? Test predictions of benchmark crystals Test predictions of benchmark crystals Prediction of experimental data Prediction of experimental data Incorporation of additional pseudopotentials Incorporation of additional pseudopotentials Cosmetics and website. Cosmetics and website.

24. Departamento de Física University of Buenos Aires Argentina Marta Ferraro Víctor Bazterra Center for High Performance Computing University of Utah Julio C. Facelli Martin Cuma