English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

We address the task of controlled generation of small molecules, whichentails finding novel molecules with desired properties under certainconstraints (e.g., similarity to a reference molecule). Here we introduceMolMIM, a probabilistic auto-encoder for small molecule drug discovery thatlearns an informative and clustered latent space. MolMIM is trained with MutualInformation Machine (MIM) learning, and provides a fixed length representationof variable length SMILES strings. Since encoder-decoder models can learnrepresentations with ``holes'' of invalid samples, here we propose a novelextension to the training procedure which promotes a dense latent space, andallows the model to sample valid molecules from random perturbations of latentcodes. We provide a thorough comparison of MolMIM to several variable-size andfixed-size encoder-decoder models, demonstrating MolMIM's superior generationas measured in terms of validity, uniqueness, and novelty. We then utilizeCMA-ES, a naive black-box and gradient free search algorithm, over MolMIM'slatent space for the task of property guided molecule optimization. We achievestate-of-the-art results in several constrained single property optimizationtasks as well as in the challenging task of multi-objective optimization,improving over previous success rate SOTA by more than 5\% . We attribute thestrong results to MolMIM's latent representation which clusters similarmolecules in the latent space, whereas CMA-ES is often used as a baselineoptimization method. We also demonstrate MolMIM to be favourable in a computelimited regime, making it an attractive model for such cases.

Improving Small Molecule Generation using Mutual Information Machine