In the Journal of Computer-Aided Molecular Design
The Alchemite™ machine learning method has been featured in a new review of massively-multitask regression models (MMRMs) for use in drug discovery. The value of such methods lies in the fact that they can be trained on large compound and assays datasets to predict bioactivity. The paper finds MMRM methods particularly suited to hit-finding, off-target, promiscuity, MoA, polypharmacology or drug-repurposing.
Citation Martin E.J., Zhu XW., Riley P. et al. Comparing massively-multitask regression algorithms for drug discovery. J Comput Aided Mol Des 40, 58 (2026). https://doi.org/10.1007/s10822-026-00761-1
Strengths of Alchemite™
The paper identified the following as key strengths of the Alchemite™ method:
- Alchemite is available via easy-to-use no-code web graphical user interfaces with the expected enterprise features and security
- Uncertainty quantification enables a user to focus on the most confident predictions
- Its “Importance Matrix” analytic highlights the inter-assay and descriptor-assay relationships
- Bayesian experimental design is directly integrated with the predictive modeling component to properly guide the next round of experiments
- Alchemite models categorical data on an equal footing to continuous/regression data.
