Method and device for partitioning a molecule
AssigneeVerseon Corporation
InventorsSachin Ahuja, David Kita, Eniko Fodor, Adityo Prakash
Life of this patent
Estimating the term…
Abstract
A method for partitioning a molecular subset is described. The partitioning method takes into account molecular structure and its manner of storage and transmission, transformations to be applied to the molecular subset and their implementation, and constraints imposed by the implementation of the partitioning method. Using this method, a molecular subset can be stored, transmitted, and processed more efficiently. The resulting efficiency makes it possible to design and run applications which require complex molecular processing, such as rational drug discovery, virtual library design, etc.
Claims58
A method for improving computational efficiency in computing measures of a biomolecule's suitability as a lead candidate for a target using a circuit having a plurality of storage units each having a maximum size of storage available, the method comprising:
- receiving, at a partitioning engine of the circuit, a molecular representation of a molecular subset of the biomolecule or the target or both, the molecular subset comprising a plurality of atoms and bonds;
- receiving, at the partitioning engine, a set of molecular transformations for all or at least one part of the molecular representation, wherein the set of molecular transformations includes at least one rotation operator;
- receiving, at the partitioning engine, a set of partitioning operators and a set of partitioning criteria, wherein the partitioning criteria includes the indication of maximum size of a storage unit of the circuit;
- partitioning, with the partitioning engine, the molecular representation into a plurality of resultant subrepresentations, wherein the partitioning depends on the set of molecular transformations, the set of partitioning operators, the set of partitioning criteria, and the maximum size of each of the plurality of storage units of the circuit, and wherein the partitioning is further based on an amount of unused storage space for storing each subrepresentation into a respective storage unit such that, by partitioning, the amount of unused storage space of the plurality of storage units of the circuit is reduced;
- and
- computing a binding affinity between at least part of the biomolecule and at least part of the target, wherein computing the binding affinity includes performing a calculation involving the molecular subset using a processor, wherein the calculation includes computing molecular transformations for respective resultant subrepresentations that resulted from partitioning the molecular representation into the plurality of resultant subrepresentations, and wherein the binding affinity corresponds to the biomolecule's suitability as a lead candidate.
A molecular processing system, including for use in improving computational efficiency in computing measures of a biomolecule's suitability as a lead candidate for a target, the molecular processing system comprising:
- a read data engine that receives a molecular representation of a molecular subset of the biomolecule or the target or both, the molecular subset including a plurality of atoms and bonds;
- one or more storage modules each having a maximum size of storage available for storing:
- a set of molecular transformations for all or at least one part of the molecular representation, wherein the set of molecular transformations includes at least one rotation operator;
- a set of partitioning operators;
- and
- a set of partitioning criteria, wherein the partitioning criteria includes a maximum size of the storage module;
- a molecule partitioning engine coupled with the read data engine and the storage modules, wherein the molecule partitioning engine is configured to partition the molecular representation into a plurality of resultant subrepresentations based on the set of molecular transformations, the set of partitioning operators the set of partitioning criteria, and the maximum size of each of the plurality of storage modules, and wherein the partitioning is further based on an amount of unused storage space for storing each subrepresentation into a respective storage module such that, by partitioning, the amount of unused storage space of the plurality of storage modules of the circuit is reduced;
- one or more molecular transform engines for transforming at least one of the resultant subrepresentations that resulted from partitioning the molecular representation into the plurality of resultant subrepresentations;
- and
- one or more molecular subset processing engines that compute binding affinity between at least part of the biomolecule and at least part of the target, wherein computing the binding affinity includes performing a calculation involving the molecular subset, the calculation using an output of the one or more molecular transform engines, and wherein the binding affinity corresponds to the biomolecule's suitability as a lead candidate.
Dependencies are inferred from the claim text.