30 Years of Molecular Field Theory Or how. days. E. Adam Kallel. June 29, Retrophin, Inc.

Size: px

Start display at page:

Download "30 Years of Molecular Field Theory Or how. days. E. Adam Kallel. June 29, Retrophin, Inc."

Theodore Knight
5 years ago
Views:

1 30 Years of Molecular Field Theory Or how good you whippersnappers have it these days. E. Adam Kallel June 29, Retrophin, Inc.

2 It hasn t been that long since we figured out what molecules were and looked like A water molecule as hook-and-eye model might have represented it. Leucippus, Democritus, Epicurus, Lucretius and Gassendi adhered to such conception. Note that the composition of water was not known before Avogadro (c. 1811). In 1803 John Dalton took the atomic weight of hydrogen, the lightest element, as unity, and determined, for example, that the ratio for nitrous anhydride was 2 to 3 which gives the formula N2O3. Interestingly, Dalton incorrectly imagined that atoms hooked together to form molecules. Later, in 1808, Dalton published his famous diagram of combined "atoms": The first proposal of the modern structure for benzene was due to Kekulé, in The cyclic nature of benzene was finally confirmed by the crystallographer Kathleen Lonsdale. Benzene presents a special problem in that, to account for all the bonds, there must be alternating double carbon bonds: 2

In 1865, August Wilhelm von Hofmann was the first to make stick-and-ball molecular models, which he used in lecture at the Royal Institution In 1874, Jacobus Henricus van 't Hoff and Joseph Achille

of a regular tetrahedron. This led to a better understanding of the three-dimensional nature of molecules. Gilbert N.

3 In 1865, August Wilhelm von Hofmann was the first to make stick-and-ball molecular models, which he used in lecture at the Royal Institution In 1874, Jacobus Henricus van 't Hoff and Joseph Achille Le Bel independently proposed that the phenomenon of optical activity could be explained by assuming that the chemical bonds between carbon atoms and their neighbors were directed towards the corners of a regular tetrahedron. This led to a better understanding of the three-dimensional nature of molecules. Gilbert N. Lewis in his famous 1916 article The Atom and the Molecule, introduced the Lewis structure to represent atoms and molecules, where dots represent electrons and lines represent covalent bonds. In this article, he developed the concept of the electron-pair bond, in which two atoms may share one to six electrons, thus forming the single electron bond, a single bond, a double bond, or a triple bond. "for his research into the nature of the chemical bond and its application to the elucidation of the structure of complex substances". 3

4 The first system for the interactive display of molecular structures was devised at MIT in the mid-1960s. Taking advantage of Project MAC (Multi-Access Computer), one of the early time-sharing mainframe computers, Cyrus Levinthal and his colleagues designed a "model-building" program to work with protein structures. This program allowed the study of short-range interaction between atoms and the "online manipulation" of molecular structures. The display terminal (nicknamed Kluge) was a monochrome oscilloscope, showing the structures in wireframe fashion. Three-dimensional effect was achieved by having the structure rotate constantly on the screen. To compensate for any ambiguity as to the actual sense of the rotation, the rate of rotation could be controlled by globe-shaped device on which the user rested his/her hand (an ancestor of today's trackball) 4

1979 Garland Marshal publishes the AAA and results in the founding of TRIPOS 1988 Richard Cramer and David Paterson publish CoMFA Tripos patents the use of PLS and grid-probe systems to visualize

5 1979 Garland Marshal publishes the AAA and results in the founding of TRIPOS 1988 Richard Cramer and David Paterson publish CoMFA Tripos patents the use of PLS and grid-probe systems to visualize molecular fields 1994 Catalyst published by Steven Teig 1994 CoMSIA published by Gerhard Klebe During the intervening period many different 3D pharmacophore modeling algorithms were developed- all were equally right or equally wrong, depending on your viewpoint Vinter founded Cresset. June 18, 2011 The last of Tripos patents expire June 30, 2017 Certara retires SYBYL 5

6 CoMFA Cramer and Milne (1979) Comparison of molecules by alignment and field generation Wold (1986) Proposes using PLS instead of PCA for overrepresented (1000 s of field non-orthogonal variables ) problem (correlate field values with activities) Cramer, Patterson and Bunce (1988) Introduced CoMFA 6

7 7

8 In the distant past of 1991, a young computational chemist faced his first industrial job. Crystal structures were few and far between. The PDB came on one CD and you had to subscribe to it. Systematic conformational searching with distance maps was the sin qua non. However, you never actually got an alignment Aligned database with charges was absolutely required Producing a CoMFA model was weeks of tedious work 8

9 3D-QSAR with CoMFA Designed to rapidly screen candidates for a particular target property Correlates changes in activity with changes in molecular properties such as biological activity Based on assumption that target property is shape- and structuredependent Requires no initial active site information Requires congeneric training series of successful compounds 1. All compounds in training series interact with same binding site in same manner 2. Interactions responsible for critical effect non-covalent and enthalpic 1. Entropic terms vary minimally between training series compounds 2. Solvent effects, diffusion/transport factors not required to derive good model 9

10 3D-QSAR with CoMFA Clustering and statistical analysis tools PLS Clustering Hierarchical, Jarvis-Patrick clustering SIMCA classification Descriptors CoMFA, CoMSIA, EVA, Model validation progressive scrambling and other tools CoMFA most predictive and interpretable models requires molecular alignment (GASP, DISCOtech, FlexS) 10

11 Charges, you say? how ever did you compute them? Population analysis of wavefunctions Mulliken population analysis Coulson's charges Natural charges CM1, CM2, CM3, CM4, and CM5 charge models Partitioning of electron density distributions Bader charges (obtained from an atoms in molecules analysis) Density fitted atomic charges Hirshfeld charges[11] Maslen's corrected Bader charges Politzer's charges Voronoi Deformation Density charges Density Derived Electrostatic and Chemical (DDEC) charges, which simultaneously reproduce the chemical states of atoms in a material and the electrostatic potential surrounding the material's electron density distribution Charges derived from dipole-dependent properties Dipole charges Dipole derivative charges, also called atomic polar tensor (APT) derived charges, or Born, Callen, or Szigeti effective charges Charges derived from electrostatic potential Chelp ChelpG (Breneman model) Merz-Singh-Kollman (also known as Merz-Kollman, or MK) Charges derived from spectroscopic data Charges from infrared intensities Charges from X-ray photoelectron spectroscopy (ESCA) Formal charges. Electronegativity-based charges 11

Electrostatics in Molecules 1,5,9-trioxo-13-azatriangulene (TOAT) Nat. Commun. 7:11560 doi: 10.1038/ncomms11560 (2016).

Crosses indicate characteristic vertices. (b) Same as a but measured with a CO tip. (c) electrostatic force field calculated from DFT.

12 Electrostatics in Molecules 1,5,9-trioxo-13-azatriangulene (TOAT) Nat. Commun. 7:11560 doi: /ncomms11560 (2016). 12 Figure 4 Determining the electrostatic field above an individual molecule. (a) High-pass filtered constant-height AFM images of a TOAT molecule on Cu(111) acquired with a Xe tip. Crosses indicate characteristic vertices. (b) Same as a but measured with a CO tip. (c) electrostatic force field calculated from DFT. (d) experimentally determined electrostatic force field obtained after subtraction of the van der Waals component from the deformation field obtained from the images shown in a and b. (e) calculated Hartree potential; (f) electrostatic potential calculated from the experimental deformation field shown in d.

13 CoMFA in the 90 Alignments were obtained by Active Analogue searching with sequential distance maps. Usually there were several families of conformations Necessitating multiple alignments. Hope the Q 2 was correct Compute charges for aligned molecules Run CoMFA 13

14 The nuisance that was finding alignments. 14

15 Alignments were difficult to find. 15

16 The results of CoMFA on the RXR ligands 16

17 RXR ligands Bexarotene and 9-Cis RA 17

18 FieldTemplater Solves the alignment in 5 minutes. 18

19 The PLS model defaults in Forge produce an excellent model in short order Q 2 = 0.62 R 2 = components 19

20 CoMFA tried to evolve PLS region focusing an attempt to heighten signal CoMFA standard scaling applies the same weight to data from each lattice point in any given field. Region focusing is an iterative procedure which refines a model by increasing the weight for those lattice points which are most pertinent to the model. This enhances the resolution and predictive power (q2; crossvalidated r2) of a subsequent PLS analysis. Technically, this corresponds to rotating the model components through a high-order space. An iterative protocol is followed: a region is defined, a CoMFA column is generated, and a PLS analysis is run. The Region Focusing routine then takes the PLS analysis and an associated CoMFA column as input, and creates a new, focused region. You are then prompted to create a new CoMFA column(s) based on that region and, therefrom, new PLS analyses. 20

21 Topomer CoMFA was the last evolutionary step Topomer CoMFA was an attempt to deal With the alignment problem. It still required congeneric series of ligands The ligands were fragmented into Topomers These topomers were aligned The model was generated as a normal CoMFA It was still fairly limited and by this time CoMFA Experiencing competition (Cresset, Inteligand, Schrodinger, etc..) 21

22 Conclusions of a sort We have come a long way We finally have the technology to fulfill the promises of 30 years ago We have gone from a handful of companies in the space to more than I can list More discernment is required now than ever before. 22

23 The Answer The Chemical Structure of a Molecule Resolved by Atomic Force Microscopy 1.Leo Gross 1,*Fabian Mohn 1, Nikolaj Moll 1, Peter Liljeroth 1, 2 Gerhard Meyer 1 Science 28 Aug 2009:Vol. 325, Issue 5944, pp DOI: /science

5.1. Hardwares, Softwares and Web server used in Molecular modeling

5. EXPERIMENTAL The tools, techniques and procedures/methods used for carrying out research work reported in this thesis have been described as follows: 5.1. Hardwares, Softwares and Web server used in