Methods and apparatus for in silico prediction of chemical reactions

Patent metadata

Field	Value
Publication number	US-2016103979-A1
Application number	US-201514854222-A
Country	US
Kind code	A1
Filing date	Sep 15, 2015
Priority date	Oct 13, 2014
Publication date	Apr 14, 2016
Grant date	—

Abstract

Provided are a method and apparatus for designing and processing a rule pipeline for in silico prediction of chemical reactions. The method includes designing a rule pipeline from at least one rule for chemical conversion and processing at least one input molecule by using the designed rule pipeline to predict a chemical reaction based on a processing result of the processing.

First claim

What is claimed is:

1 . A method for in silico prediction of chemical reactions, the method comprising:
designing, by at least one processor of an in silico prediction apparatus, a rule pipeline; processing, by the at least one processor, at least one input molecule using the designed rule pipeline; and predicting, by the at least one processor, a chemical reaction based on the at least one processed input molecule, wherein the designed rule pipeline includes an arrangement of a plurality of phases including at least one rule for chemical conversion.

2 . The method of claim 1 , wherein the designing of the rule pipeline comprises:
processing the at least one rule for chemical conversion based on at least one physiochemical attribute defined in a schema; grouping the processed at least one rule for chemical conversion into at least one group; and designing the rule pipeline for each of the at least one group.

3 . The method of claim 2 , wherein the designing of the rule pipeline for each of the at least one group comprises:
processing the plurality of rules of the one group to extract at least one rule; and arranging the at least one extracted rule, wherein the arranging of the at least one extracted rule comprises arranging the at least one phase including the at least one extracted rule, and in the arranging of the at least one phase, the at least one phase is arranged such that at least one output of a current phase is fed as at least one input of a subsequent phase.

4 . The method of claim 3 , wherein the designing of the rule pipeline for each of the at least one group further comprises connecting a plurality of extracted rules to one another via a logical operator.

5 . The method of claim 3 , wherein each of the at least one extracted rule is one of a validation rule and a transformation rule.

6 . The method of claim 3 , wherein in the designing of the rule pipeline, each phase is designed to perform one of validation and transformation based on the rule included in the each phase.

7 . The method of claim 1 , wherein in the designing of the rule pipeline, a lead phase arranged as a lead phase of the rule pipeline receives the at least one input molecule while a terminal phase arranged as a last phase generates the at least one processed molecule as an output.

8 . The method of claim 1 , wherein the designing of the rule pipeline comprises designing at least one phase for receiving, generating, or modifying associated information, and
the associated information comprises an atom of a molecule that is received as an input by a current phase and is processed in a previous phase, an atom of at least one molecule that is eligible for processing in the current phase, a result of at least one computation performed previously, at least one user parameter, and at least one user configuration, wherein the generating or modifying of the associated information is performed by at least one of adding, deleting, and overwriting the associated information.

9 . The method of claim 8 , wherein in the designing of the rule pipeline, the at least one rule pipeline is designed such that one of validation or transformation of the input molecule is performed based on the associated information in each of the at least one phase.

10 . The method of claim 8 , wherein in the designing of the rule pipeline, a phase is designed to omit processing of at least atom with respect to at least one molecule received as an input by at least one phase, and information about the at least one atom for which the processing is omitted is included in the associated information.

11 . The method of claim 1 , wherein the designing of the rule pipeline comprises designing at least one phase as an optional phase,
wherein if the optional phase fails to produce an output from a received input, the optional phase generates the received input as an output with no modification.

12 . The method of claim 1 , wherein the designing of the rule pipeline comprises designing at least one phase as a metaphase including at least one rule pipeline, and
the metaphase comprises at least one rule pipeline, and if the metaphase comprises at least two rule pipelines, the rule pipelines are arranged in parallel.

13 . The method of claim 12 , in the designing of the rule pipeline, the metaphase is designed such that the at least one input molecule is validated or transformed.

14 . The method of claim 1 , wherein in the designing of the rule pipeline, the designing of the rule pipeline is performed via a directed graph,
wherein the at least one phase receives at least one input from at least one previous phase and forwards the at least one output to at least one subsequent phase.

15 . The method of claim 14 , wherein the designing of the rule pipeline comprises grouping the plurality of phases into at least one stage,
wherein a terminal phase arranged as a last phase generates the at least one processed molecule as an output, and a lead phase arranged as a lead phase receives the at least one input molecule.

16 . The method of claim 14 , wherein in the designing of the rule pipeline, the at least one rule pipeline is designed such that a plurality of rules included in a current phase applies independently or cumulatively to at least one input received by the current phase from a plurality of previous phases.

17 . The method of claim 1 , wherein the processing of the at least one input molecule via the rule pipeline comprises:
receiving the at least one molecule as input; and sequentially processing the at least one input molecule according to at least one phase of the rule pipeline, wherein in the sequentially processing, at least one output of at least one phase from among the at least one phase is fed as an input of a subsequent phase.

18 . The method of claim 1 , wherein the predicting of a chemical reaction based on the at least one processed molecule comprises selecting one of the at least one phase to predict at least one chemical reaction based on the selected phase.

19 . An in silico prediction apparatus, comprising:
a memory; and at least one processor operatively coupled to the memory, wherein the at least one processor comprises: a rule pipeline designing unit configured to design a rule pipeline; an input receiving unit configured to receive at least one molecule to be processed; a rule pipeline processing unit configured to process the at least one molecule according to at least one phase of the rule pipeline; and a chemical reaction predicting unit configured to predict at least one chemical reaction based on at least one output according to processing based on the rule pipeline.

20 . A non-transitory computer readable recording medium having embodied thereon a program for executing the method of claim 1 .

Assignees

• Samsung Electronics Co Ltd

Inventors

• Giri Varun
• Siva Kumar Venkata Tadi
• Bhaduri Anirban
• Lee Kyusang
• Dana Saswati
• Kim Taeyong

Classifications

• G16C20/10 Primary

  Analysis or design of chemical reactions, syntheses or processes · CPC title

• G06F19/701 Primary

  Physics · mapped topic