Continuous Variational Synthesis
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.
Related
- Production of membrane-embedded Bcl-2 proteins - Use of cell-free synthesis in continuous exchange for co-translational insertion of Bcl-2 proteins in lipid bilayer nanodiscs
- STRIVE: interpretable multimodal subtype discovery in Women’s cancers using an attention-weighted joint variational autoencoder
- Continuous-wave SOFISM with SPAD array detection
- MUFASA: a continuous-time stochastic framework for realistic fluorescence microscopy simulation
- Characteristic Glycemic Patterns during Capivasertib Treatment Identified by Continuous Glucose Monitoring
BibTeX
@misc{amin2026continuous,
title = {{Continuous Variational Synthesis}},
author = {Alan N. Amin and Mattia G. Gollub and Andrei Slabodkin and Elizabeth B. Wood and Eli N. Weinstein},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.35083},
eprint = {2609.35083},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.35083},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - Continuous Variational Synthesis
AU - Alan N. Amin
AU - Mattia G. Gollub
AU - Andrei Slabodkin
AU - Elizabeth B. Wood
AU - Eli N. Weinstein
PY - 2026
DA - 2026/09/28
PB - arXiv
DO - 10.48550/arxiv.2609.35083
UR - https://arxiv.org/abs/2609.35083
AB - 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.
ER -