Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Rethinking Molecular Design: Integrating Latent Variable and Auto-Regressive Models for Goal Directed Generation

Heath Arthur-Loui, Amina Mollaysa, Michael Krauthammer

TL;DR

A hybrid model in the form of a novel regularizer that leverages the strengths of both Variational Autoencoders (VAEs) and auto-regressive models to improve validity, conditional generation, and style transfer of molecular sequences is proposed.

Abstract

De novo molecule design has become a highly active research area, advanced significantly through the use of state-of-the-art generative models. Despite these advances, several fundamental questions remain unanswered as the field increasingly focuses on more complex generative models and sophisticated molecular representations as an answer to the challenges of drug design. In this paper, we return to the simplest representation of molecules, and investigate overlooked limitations of classical generative approaches, particularly Variational Autoencoders (VAEs) and auto-regressive models. We propose a hybrid model in the form of a novel regularizer that leverages the strengths of both to improve validity, conditional generation, and style transfer of molecular sequences. Additionally, we provide an in depth discussion of overlooked assumptions of these models' behaviour.

Rethinking Molecular Design: Integrating Latent Variable and Auto-Regressive Models for Goal Directed Generation

TL;DR

A hybrid model in the form of a novel regularizer that leverages the strengths of both Variational Autoencoders (VAEs) and auto-regressive models to improve validity, conditional generation, and style transfer of molecular sequences is proposed.

Abstract

De novo molecule design has become a highly active research area, advanced significantly through the use of state-of-the-art generative models. Despite these advances, several fundamental questions remain unanswered as the field increasingly focuses on more complex generative models and sophisticated molecular representations as an answer to the challenges of drug design. In this paper, we return to the simplest representation of molecules, and investigate overlooked limitations of classical generative approaches, particularly Variational Autoencoders (VAEs) and auto-regressive models. We propose a hybrid model in the form of a novel regularizer that leverages the strengths of both to improve validity, conditional generation, and style transfer of molecular sequences. Additionally, we provide an in depth discussion of overlooked assumptions of these models' behaviour.
Paper Structure (40 sections, 7 equations, 9 figures, 13 tables)

This paper contains 40 sections, 7 equations, 9 figures, 13 tables.

Table of Contents

  1. Introduction
  2. Methods
  3. Proposed Regularizer
  4. Calibration Regularizer
  5. Reward-Based Regularizer
  6. Incorporating the Regularizer
  7. Experiments
  8. Dataset
  9. Models
  10. Evaluation
  11. Experimental Results
  12. One-Shot Decoder Models
  13. Auto-Regressive Decoder Models
  14. Conclusions
  15. Literature Review
  16. ...and 25 more sections

Figures (9)

  • Figure 1: Summary statistics for ZINC250k dataset.
  • Figure 2: Examples of successful style transfer with EXP-cVAE-TF-Pol2 model.
  • Figure 3: Style transfer generation property scores for a range of target properties sampled from EXP-cVAE model.
  • Figure 4: Style transfer generation property scores for a range of target properties sampled from EXP-cVAE-KLD model.
  • Figure 5: Style transfer generation property scores for a range of target properties sampled from EXP-cVAE-Pol2 model.
  • ...and 4 more figures