Style in Melville's Anatomies

Kelley, Wyn

doi:10.1353/lvn.2022.0011

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Unfortunately, almost all publicly available realistic regression data sets for chemical tasks are small, typically under 1000 data points; therefore, we have created an ensemble of tasks with data from diverse sources, augmenting publicly available data with millions of data points from the Terray platform. In this experiment, we compare frozen embeddings from COATI (text and point), CDDD, 10 ChemBERTa MTR, 52 ChemGPT, 65 CLAMP, 8 MegaMolBART, 64 as well as fingerprints from 2048-dimensional ECFP6 (ECFP6 2048), 31 RDKit fingerprints (RDKit FP), and RDKit 2D normalized descriptors 84 on real-world activity regression tasks. To avoid advantaging or disadvantaging embeddings downstream based on the expressive power of a regressor, we fit a linear regressor 85 for each embedding.…”

Section: ■ Resultsmentioning

confidence: 99%

COATI: Multimodal Contrastive Pretraining for Representing and Traversing Chemical Space

Kaufman,

Williams,

Underkoffler

et al. 2024

J. Chem. Inf. Model.

View full text Add to dashboard Cite

Creating a successful small molecule drug is a challenging multiparameter optimization problem in an effectively infinite space of possible molecules. Generative models have emerged as powerful tools for traversing data manifolds composed of images, sounds, and text and offer an opportunity to dramatically improve the drug discovery and design process. To create generative optimization methods that are more useful than brute-force molecular generation and filtering via virtual screening, we propose that four integrated features are necessary: large, quantitative data sets of molecular structure and activity, an invertible vector representation of realistic accessible molecules, smooth and differentiable regressors that quantify uncertainty, and algorithms to simultaneously optimize properties of interest. Over the course of 12 months, Terray Therapeutics has collected a data set of 2 billion quantitative binding measurements of small molecules to therapeutic targets, which directly motivates multiparameter generative optimization of molecules conditioned on these data. To this end, we present contrastive optimization for accelerated therapeutic inference (COATI), a pretrained, multimodal encoder-decoder model of druglike chemical space. COATI is constructed without any human biasing of features, using contrastive learning from text and 3D representations of molecules to allow for downstream use with structural models. We demonstrate that COATI possesses many of the desired properties of universal molecular embedding: fixed-dimension, invertibility, autoencoding, accurate regression, and low computation cost. Finally, we present a novel metadynamics algorithm for generative optimization using a small subset of our proprietary data collected for a model protein, carbonic anhydrase, designing molecules that satisfy the multiparameter optimization task of potency, solubility, and drug likeness. This work sets the stage for fully integrated generative molecular design and optimization for small molecules.

show abstract