Papers for

synthetic biology engineers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Amortized approach improves thompson sampling for design simulators

Amortized Thompson Sampling and Efficient Information-Directed Sampling for Simulator-Based Design

Abstract: Thompson sampling is a widely used strategy for sequential decision-making under uncertainty, offering a principled way to balance exploration and exploitation. This strategy involves maximizing a target objective under a single sample drawn from the current belief over unknown quantities. Conventionally, this optimization is performed once the belief has been inferred from data, and it becomes computationally burdensome when many Thompson samples are required, especially in continuous action spaces. In this work, we introduce an approach that amortizes this process ahead of time, before any data is collected, so that an approximation of the full Thompson sampling distribution is available immediately once observations arrive. We employ this approach to solve the multi-round setting, involving sequential data acquisition and Thompson-sample recommendation across rounds. Beyond accelerating Thompson sampling, our approach enables efficient estimation of expected information gain about the optimal action. This can be used to select more informative Thompson samples or to obtain maximally informative designs with optimization. Combining this estimate of expected information gain with immediate regret supports information-directed sampling (IDS), which can outperform Thompson sampling by explicitly balancing these two quantities. We demonstrate our approach on two simulator-based design problems in synthetic biology, where the simulators have intractable likelihoods.

Fri 23 OctMachine Learning
The gist
Choosing the best action when you're unsure can be tough, especially when you need to try many possibilities. The authors developed a method that prepares the hard work ahead of time so decisions can be made faster once new information comes in. This method also helps pick actions that teach us the most about what’s best, balancing learning and doing. They showed their approach works well in designing biological systems using complex simulators.
Open → 2610.32514v1

Encoding information on DNA using enzyme-created label patterns

Representing Information on DNA using Patterns Induced by Enzymatic Labeling

Abstract: Enzymatic DNA labeling is a powerful tool with applications in biochemistry, molecular biology, biotechnology, medical science, and genomic research. This paper contributes to the evolving field of DNA-based data storage by presenting a formal framework for modeling DNA labeling in strings, specifically tailored for data storage purposes. Our approach involves a known DNA molecule as a template for labeling, employing patterns induced by a set of designed labels to represent information. One hypothetical implementation can use CRISPR-Cas9 and gRNA reagents for labeling. Various aspects of the general labeling channel, including fixed-length labels, are explored, and upper bounds on the maximum size of the corresponding codes are given. The study includes the development of an efficient encoder-decoder pair that is proven to be optimal in terms of maximum code size under specific conditions. The codewords for this framework are codesets, i.e., each codeword is a collection of a few labels. Of special interest are codes whose codesets contain fixed-length labels. Their construction and analysis are tied to de Bruijn sequences and constrained coding.

Fri 16 OctInformation Theory
The gist
Storing information on DNA is a new way to keep data safe and dense. This paper shows how patterns made by special enzymes that label DNA can represent information like digital code. The authors created a system to model this labeling and made efficient ways to encode and decode data. They also studied the limits of how much data can be stored using this method. Their work links to special sequences that help with reliable coding.
Open → 2610.24171v1

Tango improves RNA design by combining shape and pairing rules

TANGO: RNA Topology and Geometry Co-Design

Abstract: Designing RNA sequences that satisfy target tertiary-structure objectives is an important problem in RNA engineering, with broad applications in synthetic biology and therapeutics. Existing 3D RNA design methods have made substantial progress, but they largely treat the target fold as a geometry-only objective. This is incomplete for RNA, whose functional folds are specified not only by tertiary geometry but also by secondary base-pairing topology. We therefore formulate RNA inverse design as a topology-and-geometry co-design problem and propose TANGO, a multi-objective multi-agent reinforcement learning framework for this task. TANGO adopts a staged curriculum: it first learns a topology-aware policy from secondary-structure targets, then transfers this policy to tertiary-structure targets to train a general co-design model, and finally applies frontier-guided fine-tuning to refine target-specific trade-offs among secondary-structure fidelity, tertiary-structure fidelity, and sequence diversity. Experiments show that TANGO generates diverse and novel designs, improving tertiary- and secondary-structure performance by 18.3% and 52.7%, respectively, over the strongest baseline.

Wed 14 OctMachine LearningMultiagent Systems
The gist
Designing RNA molecules to fold into specific 3D shapes is important for biology and medicine, but it’s tricky because the shapes depend on both the physical form and the chemical base pairs inside the RNA. The authors treat the problem as combining these two aspects and developed a new method called TANGO that learns to design RNA sequences by first focusing on base pairing patterns and then on the overall 3D structure. Their approach uses a multi-step learning process to balance shape accuracy, pairing rules, and diversity in designs. Experiments show that TANGO creates RNA designs that match both criteria better than previous methods.
Open → 2610.20277v1

Agentic program search finds DNA design methods that generalize well

AutoNovo: Agentic Program Search Discovers Transferable Nucleic Acid Design Algorithms

Abstract: In-silico optimization of nucleic acids relies on predictive oracles that imperfectly capture biological function, so sequences optimized against one model may exploit model-specific artifacts and fail to generalize to independent models. Recent studies have shown that not all designers generalize equally well to independent models. In this work, we use agentic program search to iteratively propose, execute, evaluate, and revise code for nucleic-acid designers. We describe the successful strategies used to prompt the search. AutoNovo designers achieved the best observed mean among the evaluated methods on 3/3 DNA-design tasks in terms of in-silico performance, and 6/6 transfer tasks in terms of independent models. The designers come from a family of discovered nucleic acid designers rather than a single universal best designer. We plan to release the resulting designers. These results are the first steps in creating nucleic acid designers that maximize cross-model generalization, and thus improving the odds that in-silico designed sequences reflect real biology.

Mon 12 OctMachine LearningArtificial Intelligence
The gist
Computers often design DNA or RNA by testing sequences with models that don’t perfectly match real biology, so designs can work on one model but fail on others. The authors used a technique where a program writes, tests, and improves code automatically to discover new ways to design DNA sequences. Their discovered design methods performed better across different prediction models than existing approaches. This helps create DNA sequences that are more likely to work well in real biological settings, not just on one computer model.
Open → 2610.16232v1

Continuous variational synthesis enables diverse designed dna production

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.

Mon 28 SeptMachine Learning
The gist
Creating large numbers of specifically designed DNA sequences was limited by how well machines could make them. The authors developed a method where they train models in a smooth, continuous way to design DNA, then convert those designs into forms that meet physical manufacturing limits. This approach helps make diverse and high-quality DNA sequences for enzymes, peptides, antibodies, and regulatory parts. The designs worked well both on computer simulations and in lab experiments.
Open → 2609.35083v1