Here is the raw dataset and codes used in TPA machine learning project of Yuming Su and Yiheng Dai. This project is supervised by Prof.Pavlo O Dral, Da Zhou and Cheng Wang in Xiamen Univerisity. environment.yml shows the anaconda packages used in this project, helped you to run this code in your own machine.
If this work is help to your scientific work, please cite our work.
How to cite
Su, Y., Dai, Y., Zeng, Y., Wei, C., Chen, Y., Ge, F., Zheng, P., Zhou, D., Dral, P. O., Wang, C., Interpretable Machine Learning of Two-Photon Absorption. Adv. Sci. 2023, 10, 2204902. https://doi.org/10.1002/advs.202204902
We think Comprehensive_Molecular_Featurizer_V2.0 is possibly helpful, because it can generate feature matrix based on Molecular fragment fingerprint(MFF), RDkit, and MFF-MOE. Especially, a python function, find_conju, could help to find the conjugated parts in one molecule. The Input shuold be mol from smiles.
The data of two-photon absorption of this study are openly available in figshare at https://doi.org/10.6084/m9.figshare.c.6264228.v2, reference number 6264228. Predictions of ML-TPA cross section can be done with open source, free package MLatom at http://mlatom.com. In addition, calculations can be performed online using the MLatom@XACS (Xiamen Atomistic Computing Suite) cloud computing service at http://XACScloud.com