Revealing the Relationship Between Publication Bias and Chemical Reactivity with Contrastive Learning

2024-02-19

AI Summary

This work identifies a pervasive publication bias in substrate scope tables within chemical literature—successful reactions are overrepresented, while failed or low-yielding experiments are rarely reported, leading to skewed reactivity perception. To address this, we propose “substrate scope contrastive learning”: leveraging published substrates as positive samples and chemically feasible yet unreported substrates as negative samples, we train a graph neural network on the CAS Content Collection™ aryl halide dataset (2010–2015, 20,798 instances) to learn discriminative molecular representations. Crucially, we formalize publication bias as a learnable chemical prior for the first time; its learned embedding space exhibits strong correlation (R² > 0.6) with established physical organic descriptors such as Hammett constants. The resulting representations significantly improve performance on downstream tasks—including yield prediction and regioselectivity classification—demonstrating that historical literature, when properly de-biased, constitutes a high-quality source of reaction knowledge.

Technology Category

• Machine Learning: Representation Learning
• Knowledge Representation and Reasoning: Preferences
• Cognitive Modeling & Cognitive Systems: Symbolic Representations

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

A synthetic method's substrate tolerance and generality are often showcased in a"substrate scope"table. However, substrate selection exhibits a frequently discussed publication bias: unsuccessful experiments or low-yielding results are rarely reported. In this work, we explore more deeply the relationship between such publication bias and chemical reactivity beyond the simple analysis of yield distributions using a novel neural network training strategy, substrate scope contrastive learning. By treating reported substrates as positive samples and non-reported substrates as negative samples, our contrastive learning strategy teaches a model to group molecules within a numerical embedding space, based on historical trends in published substrate scope tables. Training on 20,798 aryl halides in the CAS Content Collection$^{ ext{TM}}$, spanning thousands of publications from 2010-2015, we demonstrate that the learned embeddings exhibit a correlation with physical organic reactivity descriptors through both intuitive visualizations and quantitative regression analyses. Additionally, these embeddings are applicable to various reaction modeling tasks like yield prediction and regioselectivity prediction, underscoring the potential to use historical reaction data as a pre-training task. This work not only presents a chemistry-specific machine learning training strategy to learn from literature data in a new way, but also represents a unique approach to uncover trends in chemical reactivity reflected by trends in substrate selection in publications.

Problem

• Analyzes publication bias in chemical reactivity studies.
• Develops contrastive learning for substrate scope analysis.
• Links historical data to reactivity and yield prediction.

Innovation

• Contrastive learning for reactivity analysis
• Embedding molecules using historical trends
• Predicting yields and regioselectivity effectively