Abstract Details


Poster 19: CONFECT: Generating Conformations from an Expert Collection of Torsion Patterns

Christin Schärfer1, Tanja Schulz-Gasch2, Matthias Rarey1
1Center for Bioinformatics (ZBH), Bundesstr. 43, 20146 Hamburg, Germany
2F. Hoffmann-La Roche Ltd., 4070 Basel, Switzerland
Molecular conformations are used in a wide range of virtual screening applications to represent the conformational flexibility of a molecule. The underlying conformation model has a major impact on the results of these applications, which makes the generation of conformations a central task in molecular modeling. Various methods addressing the generation of conformations already exist, but most of them lack transparency and do not fully disclose how conformations are generated, thus limiting user influence on the generation process. To address these problems we developed a new conformer generator tool called CONFECT.

The generation of conformations in CONFECT is based on an expert system of torsion patterns (torsion library) which can easily be modified and extended. Torsion patterns are defined in SMARTS line notation and describe a molecular substructure that includes at least the four atoms needed to define a torsion angle. Each torsion pattern is associated with two frequency histograms and a list of torsion angles and tolerances. Histograms are generated from CSD[1] and PDB[2] data respectively. The list of torsion angles is derived from peaks in the CSD histograms, representing frequently observed torsion angles. To also take the width of a histogram peak into account, two tolerances are added to each peak torsion angle.

To generate conformations, the CONFECT algorithm starts by assigning a torsion pattern to each rotatable bond of the given molecule and then dividing the molecule into components at each rotatable bond. Conformations are constructed by adding one component after another, using the list of torsion angles from the previously assigned torsion patterns. For each flexible ring with up to eight bonds, conformations are calculated prior to the assembling step by using a ring template library combined with forcefield-based structure optimization. Evaluation of conformations during assembly is done using a simple scoring function based on absolute frequencies in the histograms derived from CSD data. Conformations containing clashes that can be resolved by slight torsion angle rotation are optimized by a stochastic hill- climbing-based minimization procedure. Rotations about torsion angles are constrained by the tolerances of assigned torsion patterns. Conformations are clustered either by using TFD[3] or rmsd to obtain the final set of conformations. Different quality levels are used to include one or both tolerances during the assembling process, resulting in a more exhaustive sampling of the conformational space of the molecule.

We compared CONFECT to Catalyst[4], ICM[5,6], Omega[7] and ConfGen[8] using a data set compiled by Perola et al.[9] and the Iridium data set[10]. In terms of reproducing the bioactive conformation, ensemble size and running time, CONFECT showed comparable results. The advantage of CONFECT is,that for the generated conformations every rotatable bond can be traced back to the underlying torsion pattern and associated list of torsion angles that led to the current torsion angle value. This, together with the option to easily modify and extend the torsion library, allows the user to directly influence the conformation generation with CONFECT.

[1] Allen, F. H. The Cambridge Structural Database: a quarter of a million crystal structures and rising. Acta Cryst. 2002, B58, 380388.
[2] Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.; Bhat, T. N.; Weissig, H.;Shindyalov, I. N.; Bourne, P. E. The Protein Data Bank. Nucleic Acids Res. 2000, 28, 235242.
[3] Schulz-Gasch, T.; Schärfer, C.; Guba, W.; Rarey, M. TFD: Torsion Fingerprints As a New Measure To Compare Small Molecule Conformations. J. Chem. Inf. Model. 2012, 52, 14991512.
[4] Catalyst; Accelrys: San Diego, CA.
[5] Abagyan, R.; Totrov, M. Biased Monte Carlo conformational searches and electrostatic calculations for peptides and proteins. J. Mol. Biol. 1994, 235, 9831002.
[6] ICM; Molsoft L.L.C.: La Jolla, CA.
[7] Omega; Openeye Scientific Software: Santa Fe, NM.
[8] Watts, K.S.; Dalal, P.; Murphy, R.B.; Sherman, W.; Friesner, R.A.; Shelly J.C. ConfGen: A Conformational Search Method for Efficient Generation of Bioactive Conformer J. Chem. Inf. Model. 2010 50 (4), 534-546
[9] Perola, E.; Charifson, P. S. Conformational Analysis of Drug-Like Molecules Bound to Proteins: An Extensive Study of Ligand Reorganization upon Binding. J. Med. Chem. 2004, 47 (10), 24992510.
[10] OpenEye Scientific Software: Iridium: A Highly Trustworthy Protein- Ligand Structure Database; http://www.eyesopen.com/iridium

Return to Programme