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 Plus annual

Presents the access of premium content as premium feature

Premium Content

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 Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Activity and property prediction models are the central workhorses in drugdiscovery and materials sciences, but currently they have to be trained orfine-tuned for new tasks. Without training or fine-tuning, scientific languagemodels could be used for such low-data tasks through their announced zero- andfew-shot capabilities. However, their predictive quality at activity predictionis lacking. In this work, we envision a novel type of activity prediction modelthat is able to adapt to new prediction tasks at inference time, viaunderstanding textual information describing the task. To this end, we proposea new architecture with separate modules for chemical and natural languageinputs, and a contrastive pre-training objective on data from large biochemicaldatabases. In extensive experiments, we show that our method CLAMP yieldsimproved predictive performance on few-shot learning benchmarks and zero-shotproblems in drug discovery. We attribute the advances of our method to themodularized architecture and to our pre-training objective.

Enhancing Activity Prediction Models in Drug Discovery with the Ability to Understand Human Language