Arris Pharmaceutical Corporation

Axys Pharmaceuticals, Inc.

David Chapman

Roger Critchlow

Ajay N Jain

Richard H Lathrop

Tomas Lozano-Perez

Thomas Glen Dietterich

Rick Lathrop

Tom Dietterich

Tomas L Perez

Machine-learning approach to modeling biological activity for molecular

A method and system for selecting molecules or molecular parts for screening. A base set of molecules on which chemical tests are to be performed is defined, and once defined, the base set may be extended by selecting molecules which are determined to maximize the diversity of the extended set. Diversity of a set of molecules is measured by reference to steric, electrostatic, and hydrogen bonding features of the set of 3D conformational shapes which each molecule in the set may take on. Each molecule is assigned a measure of incremental (added) diversity with respect to the base set of molecules on which chemical tests are to be performed. To do this, for each new molecule, a set of conformations is defined, and for each conformation, a measure of dissimilarity from each conformation of those molecules in the base set is defined. The molecule's measure of added diversity is responsive to a closest match between the any conformation of that molecule and any conformation of any molecule already in the base set. This measure of dissimilarity is determined by means of a continuous function minimization technique, such as gradient descent or simulated annealing. The base set may then be extended with the molecule whose measure of added diversity is greatest.

Three dimensional measurement of molecular diversity

Explicit representation of molecular shape of molecules is combined with neural network learning methods to provide models with high predictive ability that generalize to different chemical classes where structurally diverse molecules exhibiting similar surface characteristics are treated as similar. A new machine-learning methodology that can accept multiple representations of objects and construct models that predict characteristics of those objects. An extension of this methodology can be applied in cases where the representations of the objects are determined by a set of adjustable parameters. An iterative process applies intermediate models to generate new representations of the objects by adjusting said parameters and repeatedly retrains the models to obtain better predictive models. This method can be applied to molecules because each molecule can have many orientations and conformations (representations) that are determined by a set of translation, rotation and torsion angle parameters.

Growing community of inventors

San Francisco, CA, United States of America

David Chapman