# US 12,191,003 B2 · Real-time prediction of chemical properties through combining calculated, structured and unstructured data at large scale > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,191,003 B2 - Document ID: US12191003B2 - Status: Granted - Kind: B2 - Application number: 16/148,159 - Filed: 2018-10-01 - Published: 2020-04-02 - Granted: 2025-01-07 - Earliest claimed: 2018-10-01 - Assignees: International Business Machines Corporation - Applicants: International Business Machines Corporation - Inventors: Richard L. Martin; Sheng Hua Bao - Web page: https://patentgrep.com/patent/US12191003B2 ## CPC - G16C 20/30: Prediction of properties of chemical compounds, compositions or mixtures - G16C 20/70: Machine learning, data mining or chemometrics - G16C: COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE ## Abstract A mechanism is provided in a data processing system comprising at least one processor and a memory comprising instructions which, when executed by the at least one processor, causes the at least one processor to implement a real-time prediction engine for real-time predication of chemical properties through combining calculated, structured, and unstructured data at large scale. Offline components executing within the real-time prediction engine store a computational representation for each of a plurality of chemical structures in a unified storage. Each computational representation maps a respective chemical structure to a vector of calculated chemical structure features and properties, unstructured chemical features and properties, and structured chemical features and properties. The offline components train a computational real-time predictive model based on the computational representations. A user interface executing within the real-time prediction engine receives a request specifying one or more chemical compounds. An analytics jobs manager executing within the real-time prediction engine predicts one or more properties of the one or more chemical compounds using the computational real-time predictive model. The analytics jobs manager outputs the one or more properties of the one or more chemical compounds to the user interface. ## Claims ### Claim 1 A method, in a data processing system comprising at least one processor and a memory comprising instructions which, when executed by the at least one processor, causes the at least one processor to implement a real-time prediction engine for real-time prediction of chemical properties through combining calculated, structured, and unstructured data at large scale, the method comprising: inputting, for each of a plurality of chemical structures, into a chemical information processor, unstructured chemical features and properties extracted, by a natural language processing job server, from one or more unstructured chemical information sources and structured chemical features and properties extracted, by a structured data processor, from structured chemical information sources; calculating, by the chemical information processor, for each of a plurality of chemical structures, calculated chemical structure features and properties based on the unstructured chemical features and properties, and the structured chemical features and properties; storing, by offline components executing within the real-time prediction engine, a computational representation for each of the plurality of chemical structures in a unified storage, wherein each computational representation maps a respective chemical structure to a vector of corresponding calculated chemical structure features and properties, corresponding unstructured chemical features and properties, and corresponding structured chemical features and properties; training, by the offline components using a machine learning training operation, a computational real-time predictive model based on the computational representations as inputs to the computational real-time predictive model, wherein the computational real-time predictive model is trained to predict properties based on an input chemical compound; receiving, by a user interface executing within the real-time prediction engine, a request specifying one or more chemical compounds; predicting, by an analytics jobs manager executing within the real-time prediction engine, one or more properties of the one or more chemical compounds using the computational real-time predictive model; and outputting, by the analytics jobs manager, the one or more properties of the one or more chemical compounds to the user interface, wherein the computational real-time predictive model comprises a machine learning model, a deep learning model, or a neural network. ### Claim 2 (depends on 1) The method of claim 1, further comprising processing, by the offline components, unstructured data describing the plurality of chemical structures to extract unstructured chemical features and properties. ### Claim 3 (depends on 2) The method of claim 2, wherein the unstructured chemical features and properties comprise generic names, systematic names, adjacent or co-occurring words or phrases, relationships between the chemical structure and words or phrases, assertion of attributes, or assertion of effects of the chemical structure. ### Claim 4 (depends on 1) The method of claim 1, further comprising processing, by the offline components, structured data describing the plurality of chemicals to extract structured chemical features and properties. ### Claim 5 (depends on 4) The method of claim 4, wherein the structured chemical features and properties comprise molecular weight or IC50 values. ### Claim 6 (depends on 1) The method of claim 1, further comprising generating, by the offline components for each of the plurality of chemicals, chemical structure features based on the unstructured data and the structured data. ### Claim 7 (depends on 6) The method of claim 6, wherein the chemical structure features comprise an atomistic structure of the chemical structure or a two-or three-dimensional representation, or a two-dimensional chemical graph or connection table describing how individual constituent atoms of the chemical structure are chemically bonded. ### Claim 8 (depends on 1) The method of claim 1, wherein predicting the one or more properties of the one or more chemical compounds comprises sending by an analytics job server a plurality of real-time predictive analytics jobs to a plurality of analytics job worker components. ### Claim 9 (depends on 1) The method of claim 1, wherein predicting the one or more properties of the one or more chemical compounds comprises: identifying chemical structures similar to the one or more chemical compounds; comparing the one or more chemical compounds to the identified similar chemical structures; and using the computational real-time predictive model to predict the one or more properties of the one or more chemical compounds based on results of the comparison. ### Claim 10 A computer program product comprising a non-transitory computer readable medium having a computer readable program stored therein, wherein the computer readable program, when executed on a computing device, causes the computing device to implement a real-time prediction engine for real-time prediction of chemical properties through combining calculated, structured, and unstructured data at large scale, wherein the computer readable program causes the computing device to: input, for each of a plurality of chemical structures, into a chemical information processor, unstructured chemical features and properties extracted, by a natural language processing job server, from one or more unstructured chemical information sources and structured chemical features and properties extracted, by a structured data processor, from structured chemical information sources; calculate, by the chemical information processor, for each of a plurality of chemical structures, calculated chemical structure features and properties based on the unstructured chemical features and properties, and the structured chemical features and properties; store, by offline components executing within the real-time prediction engine, a computational representation for each of the plurality of chemical structures in a unified storage, wherein each computational representation maps a respective chemical structure to a vector of corresponding calculated chemical structure features and properties, corresponding unstructured chemical features and properties, and corresponding structured chemical features and properties; train, by the offline components using a machine learning training operation, a computational real-time predictive model based on the computational representations as inputs to the computational real-time predictive model, wherein the computational real-time predictive model is trained to predict properties based on an input chemical compound; receive, by a user interface executing within the real-time prediction engine, a request specifying one or more chemical compounds; predict, by an analytics jobs manager executing within the real-time prediction engine, one or more properties of the one or more chemical compounds using the computational real-time predictive model; and output, by the analytics jobs manager, the one or more properties of the one or more chemical compounds to the user interface, wherein the computational real-time predictive model comprises a machine learning model, a deep learning model, or a neural network. ### Claim 11 (depends on 10) The computer program product of claim 10, wherein the computer readable program causes the computing device to process, by the offline components, unstructured data describing the plurality of chemical structures to extract unstructured chemical features and properties. ### Claim 12 (depends on 11) The computer program product of claim 11, wherein the unstructured chemical features and properties comprise generic names, systematic names, adjacent or co-occurring words or phrases, relationships between the chemical structure and words or phrases, assertion of attributes, or assertion of effects of the chemical structure. ### Claim 13 (depends on 10) The computer program product of claim 10, wherein the computer readable program causes the computing device to process, by the offline components, structured data describing the plurality of chemicals to extract structured chemical features and properties. ### Claim 14 (depends on 13) The computer program product of claim 13, wherein the structured chemical features and properties comprise molecular weight or IC50 values. ### Claim 15 (depends on 10) The computer program product of claim 10, wherein the computer readable program causes the computing device to generate, by the offline components for each of the plurality of chemicals, chemical structure features based on the unstructured data and the structured data. ### Claim 16 (depends on 15) The computer program product of claim 15, wherein the chemical structure features comprise an atomistic structure of the chemical structure or a two-or three-dimensional representation, or a two-dimensional chemical graph or connection table describing how individual constituent atoms of the chemical structure are chemically bonded. ### Claim 17 (depends on 10) The computer program product of claim 10, wherein predicting the one or more properties of the one or more chemical compounds comprises sending by an analytics job server a plurality of real-time predictive analytics jobs to a plurality of analytics job worker components. ### Claim 18 An apparatus comprising: a processor; and a memory coupled to the processor, wherein the memory comprises instructions which, when executed by the processor, cause the processor to implement a real-time prediction engine for real-time prediction of chemical properties through combining calculated, structured, and unstructured data at large scale, wherein the instructions cause the processor to: input, for each of a plurality of chemical structures, into a chemical information processor, unstructured chemical features and properties extracted, by a natural language processing job server, from one or more unstructured chemical information sources and structured chemical features and properties extracted, by a structured data processor, from structured chemical information sources; calculate, by the chemical information processor, for each of a plurality of chemical structures, calculated chemical structure features and properties based on the unstructured chemical features and properties, and the structured chemical features and properties; store, by offline components executing within the real-time prediction engine, a computational representation for each of the plurality of chemical structures in a unified storage, wherein each computational representation maps a respective chemical structure to a vector of corresponding calculated chemical structure features and properties, corresponding unstructured chemical features and properties, and corresponding structured chemical features and properties; train, by the offline components using a machine learning training operation, a computational real-time predictive model based on the computational representations as inputs to the computational real-time predictive model, wherein the computational real-time predictive model is trained to predict properties based on an input chemical compound; receive, by a user interface executing within the real-time prediction engine, a request specifying one or more chemical compounds; predict, by an analytics jobs manager executing within the real-time prediction engine, one or more properties of the one or more chemical compounds using the computational real-time predictive model; and output, by the analytics jobs manager, the one or more properties of the one or more chemical compounds to the user interface, wherein the computational real-time predictive model comprises a machine learning model, a deep learning model, or a neural network. ### Claim 19 (depends on 1) The method of claim 1, wherein the calculated chemical structure features and properties comprises at least one of chemical functional groups, individual smallest ring systems, and fused ring systems. ### Claim 20 (depends on 1) The method of claim 1, wherein the calculated chemical structure features and properties comprise an atomic structure specified as a three-dimensional representation in a connection table that describes how individual constituent atoms are chemically bonded, and simulated properties and attributes of the atomic structure.