This repository contains a model which predicts reactivity ratios between monomer pairs in a radical copolymerization solely based on their chemical structures. More details are described in our paper. The user interface is also available at PolymatAI.
To make predictions use "ANN_Prediction.py" file. It uses the "ANN_model.h5" file to get all the weights of the network and then predicts the reactivity ratio values based on the given inputs. Entries must be in the form of the SMILES* string. The training code is also provided in which the features are generated based on fingerprints and the model is trained based on them. All the data can be found in Polymer Handbook; Brandrup et al.; 4th edition.; Wiley, 1999. (John Wiley & Sons, Inc. holds the copyright of this database).
*The simplified molecular-input line-entry system (SMILES) is a specification in the form of a line notation for describing the structure of chemical species using short ASCII strings. You can find the SMILES string of a monomer by entering its "Structure Identifier" (such as name, CAS number, etc.) and selecting the "convert to" option as "SMILES", then clicking on submit, at Here.
Required packages and thier versions are listed in the requirements.
This repository is licensed under CC BY-NC 4.0. For more information please refer to the license section.