sources/header.md

Predict the MIC of compounds against pathogenic bacteria

Predictions are from an AI model trained on the wild-type accumulator subset of the SPARK dataset, available to browse here.

Predictions are given in micromolar (µM) and µg/mL. You can optionally have uncertainty scores calculated. These can take a few minutes, so please be patient.

This model was generated using our Duvida framework, as a result of hyperparameter searches and selecting the model that performs best on unseen test data (from a scaffold split). Duvida also allows the calculation of uncertainty metrics based on training data.

Available species for prediction are:

• Acinetobacter baumannii
• Brucella abortus
• Escherichia coli
• Francisella tularensis
• Klebsiella pneumoniae
• Pseudomonas aeruginosa
• Staphylococcus aureus
• Streptococcus pneumoniae
• Yersinia enterocolitica
• Yersinia pestis

Click on the links above for training details, model configurations, and evaluation metrics.