Continuous Variational Synthesis

2026-09-28

AI Summary

This study addresses the challenge that DNA synthesis is constrained by chemical requirements necessitating parameter discretization, which impedes effective model pretraining and fine-tuning. To overcome this bottleneck in discrete-space optimization, this work proposes a "free" variational synthesis framework that trains a variational autoencoder in continuous space and satisfies hardware constraints through post-training quantization. The proposed method significantly improves the quality-diversity Pareto front of sequence generation and successfully designs enzymes, peptides, and antibodies that strictly satisfy predefined reward criteria. Furthermore, in vitro experimental performance demonstrates high concordance with computational predictions, validating the practical efficacy of the framework for biological sequence design.

Abstract

Biological machine learning was long bottlenecked by the ability to synthesize designed DNA. Variational synthesis models control chemical reactions to physically manufacture quadrillions of designed sequences in DNA. However, training these generative models is challenging: constraints on chemical synthesis can force many parameters into a discrete space, limiting the ability to pre-train and fine-tune. In this article we train ``free''variational synthesis models using stochastic gradient descent in continuous space, and then discretize with post-training quantization to impose hardware and wetware constraints. This enables variational synthesis models to satisfy stringent reward criteria, while still synthesizing diverse designs, achieving a strictly dominating quality-diversity Pareto frontier. We demonstrate by training variational synthesis models of enzymes, peptides, antibody CDRH3s, and regulatory DNA elements. In silico performance is maintained in vitro.

Problem

• Variational synthesis
• Biological machine learning
• DNA synthesis
• Discrete space constraints
• Generative model training

Innovation

• Variational Synthesis
• Continuous Space Optimization
• Post-training Quantization
• Quality-Diversity Pareto Frontier
• Biological Machine Learning