Automated transitive read-behind analysis in big data toxicology

A computer-readable storage medium having computer-readable program code embodied therewith, the computer-readable program code executable by one or more computer processors to perform an operation comprising:

• receiving an indication of a proposed chemical composition;
• generating a predicted toxicity score for the proposed chemical composition by performing, by operation of one or more computer processors, a probabilistic read-across analysis on the proposed chemical composition, wherein performing the probabilistic read-across analysis comprises:
• identifying, based on a knowledge graph, at least one similar composition that is structurally similar to the proposed chemical composition, wherein the knowledge graph comprises:
• a plurality of nodes, each respective node corresponding to a respective chemical composition, and
• a plurality of edges indicating learned structural similarity among chemical compositions, wherein at least one of the plurality of edges comprises a strength defined based on regulatory similarity, with respect to at least one regulatory body, between corresponding chemical compositions;
• identifying a set of chemical reactions that include the at least one similar composition;
• determining one or more products of the identified set of chemical reactions;
• and
• determining a toxicity of at least one of the one or more products;
• and
• returning the predicted toxicity score, wherein the predicted toxicity score provides an understanding regarding potential risks of the proposed chemical composition and can be used to ensure that the proposed chemical composition will not be potentially dangerous or toxic.

A system comprising:

• one or more computer processors;
• and
• a memory containing a program which when executed by the one or more computer processors performs an operation, the operation comprising:
• receiving an indication of a proposed chemical composition;
• generating a predicted toxicity score for the proposed chemical composition by performing a probabilistic read-across analysis on the proposed chemical composition, wherein performing the probabilistic read-across analysis comprises:
• identifying, based on a knowledge graph, at least one similar composition that is structurally similar to the proposed chemical composition, wherein the knowledge graph comprises:
• a plurality of nodes, each respective node corresponding to a respective chemical composition, and
• a plurality of edges indicating learned structural similarity among chemical compositions, wherein at least one of the plurality of edges comprises a strength defined based on regulatory similarity, with respect to at least one regulatory body, between corresponding chemical compositions;
• identifying a set of chemical reactions that include the at least one similar composition;
• determining one or more products of the identified set of chemical reactions;
• and
• determining a toxicity of at least one of the one or more products;
• and
• returning the predicted toxicity score, wherein the predicted toxicity score provides an understanding regarding potential risks of the proposed chemical composition and can be used to ensure that the proposed chemical composition will not be potentially dangerous or toxic.