INTRODUCTION

Recursion presented multiple works of research at workshops associated with NeurIPS in 2021 and 2022. The Neural Information Processing Systems Foundation (NeurIPS) is a nonprofit fostering the exchange of artificial intelligence and machine learning research advances by hosting an annual interdisciplinary academic conference.

BIOLOGICAL CARTOGRAPHY

Building and Benchmarking Representations of Life

Presented at the Learning Meaningful Representations of Life (LMRL) Workshop at NeurIPS 2022

The continued scaling of genetic perturbation technologies combined with high-dimensional assays (microscopy and RNA-sequencing) has enabled genome-scale reverse-genetics experiments that go beyond single-endpoint measurements of growth or lethality. Datasets emerging from these experiments can be combined to construct “maps of biology”, in which perturbation readouts are placed in unified, relatable embedding spaces to capture known biological relationships and discover new ones. Construction of maps involves many technical choices in both experimental and computational protocols, motivating the design of benchmark procedures by which to evaluate map quality in a systematic, unbiased manner. 

In this work, we propose a framework for the steps involved in map building and demonstrate key classes of benchmarks to assess the quality of a map. We describe univariate benchmarks assessing perturbation quality and multivariate benchmarks assessing recovery of known biological relationships from large-scale public data sources. We demonstrate the application and interpretation of these benchmarks through example maps of scRNA-seq and phenomic imaging data.

Read the PaperDownload the Poster

Watch the Presentation

MULTI-OBJECTIVE GFLOWNETS

Multi-Objective Generative Flow Networks

Presented at the AI for Accelerated Materials Design (AI4Mat) Workshop at NeurIPS 2022

In many applications of machine learning, like drug discovery and material design, the goal is to generate candidates that simultaneously maximize a set of objectives. As these objectives are often conflicting, there is no single candidate that simultaneously maximizes all objectives, but rather a set of Pareto-optimal candidates where one objective cannot be improved without worsening another. Moreover, in practice, these objectives are often under-specified, making the diversity of candidates a key consideration. The existing multi-objective optimization methods focus predominantly on covering the Pareto front, failing to capture diversity in the space of candidates. Motivated by the success of GFlowNets for generation of diverse candidates in a single objective setting, in this paper we consider Multi-Objective GFlowNets (MOGFNs). MOGFNs consist of a novel Conditional GFlowNet which models a family of single-objective sub-problems derived by decomposing the multi-objective optimization problem. Our work is the first to empirically demonstrate conditional GFlowNets. Through a series of experiments on synthetic and benchmark tasks, we empirically demonstrate that MOGFNs outperform existing methods in terms of Hypervolume, R2-distance and candidate diversity. We also demonstrate the effectiveness of MOGFNs over existing methods in active learning settings. Finally, we supplement our empirical results with a careful analysis of each component of MOGFNs.

Read the Paper
MolE MOLECULAR FOUNDATION

A Molecular Foundation Model for Drug Discovery

Presented at the Critical Assessment of Molecular Machine Learning (ML4Molecules) and the Learning Meaningful Representations of Life (LMRL) Workshops at NeurIPS 2022

Models that accurately predict properties based on chemical structure are valuable tools in drug discovery. However, for many properties, public and private training sets are typically small, and it is difficult for the models to generalize well outside of the training data. Recently, large language models have addressed this problem by using self-supervised pretraining on large unlabeled datasets, followed by fine-tuning on smaller, labeled datasets. In this paper, we report MolE, a molecular foundation model that adapts the DeBERTa architecture to be used on molecular graphs together with a two-step pretraining strategy. The first step of pretraining is a self-supervised approach focused on learning chemical structures, and the second step is a massive multi-task approach to learn biological information. We show that fine-tuning pretrained MolE achieves state-of-the-art results on 9 of the 22 ADMET tasks included in the Therapeutic Data Commons.

Read the PaperDownload the Poster

Watch the Presentation

MAPPING BIOLOGY WITH A UNIFIED REPRESENTATION SPACE

Mapping Biology With a Unified Representation Space for Genomic and Chemical Perturbations to Enable Accelerated Drug Discovery

Presented at the Learning Meaningful Representations of Life (LMRL) Workshop at NeurIPS 2021

Biology is massively complex and highly networked, and as a result, modeling diseases and identifying safe and effective drugs to correct them is an enormous challenge. More than 90% of drugs that have been advanced to clinical trials fail before making it to the market. Over the last decade, we have witnessed exponential progress in diverse technical fields including genome editing, synthetic biology, robotics, automation, machine learning, artificial intelligence, and high-performance computation. Here we present Recursion’s application of these advancements to improve the scale and efficiency of drug discovery. We capture high-dimensional phenotypic readouts from human cells at a massive scale: 18,000 genes modeled by CRISPR knockout and 250,000 compound perturbations with a large number of replicates in multiple cell types. This enables us to unravel complex biological patterns and predict relationships between genes and compounds.Here, we demonstrate the feasibility of learning a representation of a broad range of biology, unified across genetic and chemical perturbations, that demonstrates power by recapitulating known pathways of genes and known drug mechanisms of action in diverse areas of biological function. Further, we show that the same learned representations can be used to direct search through chemical space to initiate discovery and structural optimization of new medicines.

Download the PosterWatch the Presentation
Stay informed about our COVID-19 study
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
SPONSORED BY: