Creation of new chemical compounds having desired properties using accumulated chemical data to construct a new chemical structure for synthesis
A computer implemented method of generating new chemical compounds is provided. The method includes preparing a data-driven substructure feature vector for each of a plurality of chemical compounds for which a chemical or physical property is known. The method further includes preparing a predefined component feature vector, creating a regression model to predict a target value for the chemical or physical property, and performing a search algorithm to identify substructure features that affect the target value for the chemical or physical property. The method further includes generating a candidate structure having the target value for the chemical or physical property, and synthesizing the candidate structure.
1. A computer implemented method of generating new chemical compounds, comprising:
preparing a data-driven substructure feature vector representing the atoms and bonds of a chemical structure for each of a plurality of chemical compounds for which a chemical or physical property is known;
preparing a predefined component feature vector representing backbone components, atoms, and chemical bonds;
creating a regression model to predict a target value for the chemical or physical property;
performing a search algorithm to identify substructure features that affect the target value for the chemical or physical property;
generating a candidate structure having the target value for the chemical or physical property; and
providing to a user the generated candidate structure for synthesizing the candidate structure.
2. The computer implemented method of claim 1 , further comprising testing a synthesized candidate structure to determine the actual value for the chemical or physical property.
3. The computer implemented method of claim 1 , further comprising receiving input from a user, wherein the input is a selection of one chemical or physical property and the target value for the selected property.
4. The computer implemented method of claim 1 , wherein the data-driven substructure feature vector uses simplified molecular-input line-entry system (SMILES) grammar to represent the plurality of chemical compounds, and the predefined component feature vector uses SMILES grammar to represent the predefined chemical substructures.
5. The computer implemented method of claim 1 , wherein the regression model is a kernel ridge regression.
6. The computer implemented method of claim 1 , wherein the search algorithm utilizes a particle swarm optimization algorithm.
7. The computer implemented method of claim 1 , wherein the candidate structure is generated using backbone structuring, atomistic detailing, and bond detailing.
8. The computer implemented method of claim 7 , wherein generated candidate structures that do not include specific substructures are eliminated to avoid an exponential increase in generated candidate structures.
9. A non-transitory article of manufacture tangibly embodying a computer readable program which when executed causes a computer to perform the steps of claim 1 .
10. A computer implemented method of generating new chemical compounds, comprising:
receiving input from a user selecting a property from a list of chemical and physical properties;
receiving input of a target value for the selected property;
automatically preparing a data-driven substructure feature vector representing the atoms and bonds of a chemical structure for each of a plurality of chemical compounds from a data set of chemical compounds for which the selected property is known;
preparing a predefined component feature vector from backbone information, atomistic information, and bonding information;
creating a regression model to predict a resulting value for the selected property;
perform a search algorithm to identify substructure features that affect the resulting value for the selected property;
generating a candidate structure having the target value for the selected property;
synthesizing the candidate structure; and
testing the synthesized candidate structure to determine the actual value of the selected property.
11. The computer method of claim 10 , wherein the data-driven substructure feature vector uses simplified molecular-input line-entry system (SMILES) strings to represent the plurality of chemical compounds.
12. The computer method of claim 11 , wherein the candidate structure is generated by combining the information in the data-driven substructure feature vectors and information in the predefined component feature vector.
13. The computer method of claim 12 , wherein the candidate structure is generated by concatenating portions of the SMILES strings of the data-driven substructure feature vectors and the predefined component feature vector.
14. The computer method of claim 13 , wherein the candidate structure is modified based on regularization.
15. The computer method of claim 14 , wherein the number of candidate structures is maintained below a maximum by filtering proposed candidate structures using the data-driven substructures.
16. A non-transitory computer readable storage medium comprising a computer readable program for generating new chemical compounds, wherein the computer readable program when executed on a computer causes the computer to perform the steps of:
preparing a data-driven substructure feature vector representing the atoms and bonds of a chemical structure for each of a plurality of chemical compounds for which a chemical or physical property is known;
preparing a predefined component feature vector representing backbone components, atoms, and chemical bonds;
creating a regression model to predict a target value for the chemical or physical property;
performing a search algorithm to identify substructure features that affect the target value for the chemical or physical property;
generating a candidate structure having the target value for the chemical or physical property; and
providing to a user the generated candidate structure for synthesizing the candidate structure.
17. The non-transitory computer readable storage medium of claim 16 , further comprising receiving input from a user, wherein the input is a selection of one chemical or physical property.
18. The non-transitory computer readable storage medium of claim 16 , wherein the search algorithm utilizes a particle swarm optimization algorithm.
19. The non-transitory computer readable storage medium of claim 16 , wherein the regression model is a kernel ridge regression.
20. The non-transitory computer readable storage medium of claim 16 , wherein the predefined component feature vector is based on the number of backbone components, the number of atomistic components, and the number of bond components.