Prediction of Antibiotic Activity against Burkholderia cenocepacia Using a Machine Learning Model
Bibliographic record
Abstract
A fundamental challenge in antibiotic discovery is finding new bioactive compound classes. Due to the longer timeframe and higher cost associated with conventional approaches, it has become imperative to adopt alternative antibiotic discovery paradigms. Advances in computational processing capacity enabled expanding chemical space. The expanded space allowed generation of vast, chemically diverse virtual compound libraries containing billions of compounds. In this study, we exploited the machine learning (ML) model’s ability to make predictive models and applied it to predict growth inhibitory activity in chemical scaffolds outside the training dataset. We employed a Directed-Message Passing Neural Network (D-MPNN) approach to train binary classification and regression ML models on a high-throughput screening dataset performed against Burkholderia cenocepacia previously in our laboratory. The D-MPNN belongs to Spatial-based Convolutional Graph Neural Networks (ConvGNNs), an end-to-end neural network that generates the graph representation of a molecule after iterative message passing process through molecular bonds. To avoid over-fitting and enhance the accuracy of the prediction, we additionally fed the model with 200 global molecular descriptors. The model was further optimized using Bayesian hyperparameter optimization and ensembling. The trained model attained a receiver operating characteristic curve-area under the curve (ROC-AUC) of 0.823. As a proof of principle, we employed the trained ML model to predict the bioactivity of 1,615 FDA-approved compounds and tested the bioactivity of the top 100 ranked compounds in vitro. We found 17 growth-inhibitory compounds with a linear correlation between the predicted rank and the activity. This work highlights the application of ML approaches to rapidly explore chemically diverse, ultra-large compound libraries and discern potential compounds in an inexpensive fashion, thus increasing the chance to discover early lead compounds.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".