Prediction of Cytochrome P450 Inhibition Using a Deep Learning Approach and Substructure Pattern Recognition by Zhaoyang Chen (1449478)

“…The models can be used as a powerful tool for the prediction of CYP450 inhibitors, and the SAs should provide useful information for the mechanisms of Cytochrome P450 inhibition.…”

Schematic showing how the tracking model uses the inputs to make predictions. by Owen M. O’Connor (11965460)

Performance metrics for ProSmith and previously published methods for DTA prediction on the random split of the Davis dataset. by Alexander Kroll (9304225)

Subjects:

ProSmith outperforms the ESP model in the prediction of enzyme-substrate pairs especially for molecules with limited representation in the test data. by Alexander Kroll (9304225)

Subjects:

Performance metrics for all three trained gradient boosting models with different input vectors and for their combined weighted mean prediction. by Alexander Kroll (9304225)

Subjects:

The optimal ProSmith models combine predictions based on the multimodal Transformer Network with predictions based on separate numerical representations of proteins and small molec... by Alexander Kroll (9304225)

Subjects:

Performance metrics for ProSmith and previously published methods for DTA prediction for different splitting scenarios. by Alexander Kroll (9304225)

Subjects:

For accurate DTA predictions, ProSmith requires training on identical drugs but not on similar proteins. by Alexander Kroll (9304225)

Subjects:

Replacing the final prediction layer of end-to-end trained Transformer Networks with gradient boosting models improves performance. by Alexander Kroll (9304225)

Subjects:

Coefficient of determination <i>R</i>² of ProSmith for different splitting scenarios of the Davis dataset. by Alexander Kroll (9304225)

Subjects:

We divided the ESP test set into subsets according to the number of positive data points with the same small molecule in the training set. by Alexander Kroll (9304225)

Subjects:

Performance metrics for ProSmith and ESP for the prediction of enzyme-substrate pairs. by Alexander Kroll (9304225)

Subjects:

We divided the ESP test set into subsets according to the maximal sequence identity of an enzyme compared to all training enzymes. by Alexander Kroll (9304225)

Subjects:

Drug-target affinity predictions improve with increasing drug similarity score compared to the training drugs. by Alexander Kroll (9304225)

Subjects:

Performance metrics of ProSmith and previously published methods for the prediction of Michaelis constants <i>K</i>_M. by Alexander Kroll (9304225)

Subjects:

ProSmith outperforms previous models in the prediction of Michaelis constants <i>K</i>_M especially for enzymes not highly similar to proteins in the training set. by Alexander Kroll (9304225)

Subjects:

ProSmith model overview. by Alexander Kroll (9304225)

Subjects:

Hyperparameters of the ProSmith Transformer Network. by Alexander Kroll (9304225)

Subjects:

Enzyme-substrate pair predictions improve with increasing substrate similarity compared to training substrates. by Alexander Kroll (9304225)

Subjects:

Hyperparameters of the ProSmith gradient boosting models. by Alexander Kroll (9304225)

Subjects: