Using Graph Neural Networks to Predict Positions of the Absorption Maxima of a Number of Dyes

Results are presented from developing a model for accurately predicting the wavelength of the absorption maximum of boron(III) dipyrromethenates (BODIPYs). The model is based on a graph neural network (GNN) and includes data for >2500 dyes of various natures. Statistical parameters of the model (MAE and R²) are 4 nm and 0.99 for the training set and 13.5 nm and 0.87 for the testing set. The developed model is available to the public in the GitHub repository (https://github.com/lukanov-9b/Abs_model.git).