ADMET-AI: A machine learning ADMET platform for evaluation of large-scale chemical libraries: Chemprop was trained on 41 absorption, distribution, metabolism, excretion, and toxicity …

This repository contains benchmarking scripts and data for Chemprop, a message passing neural network for molecular property prediction, as described in the paper Chemprop: Machine …

As datapoints are sampled from the dataset, the MolGraph data structures are generated on-the-fly, which requires featurization of the molecular graphs. Giving the dataloader multiple …

Aug 17, 2024 · @Rhys-McAlister please note that chemprop v2 is compatible with pytorch lightning, so you can use their Trainer class and avoid manually moving data on/off GPU. We …

Oct 2, 2024 · No, chemprop does not implement any data augmentation methods. They must be implemented manually.

Dec 4, 2024 · I have tried to train with the data set supplied with chemprop as a test case as well as with my own data set. I always run into this error. So far I have tried to define …

@sukrit2013 could you check whether the same target names are used in both 23Jan2023_train_reg.csv and data_test_only_SMILES.csv? In order to perform uncertainty …

Sep 21, 2022 · So, I went ahead and tried to generate a MoleculeDataset object using the utility method get_data_from_smiles available in chemprop/data/utils.py. But I see that the method …

Chemprop's splitting function implements our preferred method of data splitting, which is random replication. It's also possible to add your own custom cross-validation splitter, such as one of …

Jul 6, 2025 · Given my limited data, it's critical to include every datapoint in training the final production model. Is there a recommended workaround or an easy solution for this issue?