# Fast differentiable DNA and protein sequence optimization for molecular design

arXiv: https://arxiv.org/abs/2005.11275 · HTML: https://mlsanity.com/paper/2005.11275/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 22 May 2020.

## Built on by (most cited first, 11 of 11)

- [Regression Transformer: Concurrent sequence regression and generation for molecular language modeling](https://mlsanity.com/paper/2202.01338/) (2022)
- [Bridging Model-Based Optimization and Generative Modeling via Conservative Fine-Tuning of Diffusion Models](https://mlsanity.com/paper/2405.19673/) (2024)
- [Regulatory DNA sequence Design with Reinforcement Learning](https://mlsanity.com/paper/2503.07981/) (2025)
- [Ctrl-DNA: Controllable Cell-Type-Specific Regulatory DNA Design via Constrained RL](https://mlsanity.com/paper/2505.20578/) (2025)
- [ReLSO: A Transformer-based Model for Latent Space Optimization and Generation of Proteins](https://mlsanity.com/paper/2201.09948/) (2022)
- [Applying language models to algebraic topology: generating simplicial cycles using multi-labeling in Wu's formula](https://mlsanity.com/paper/2306.16951/) (2023)
- [Using GNN property predictors as molecule generators](https://mlsanity.com/paper/2406.03278/) (2024)
- [Semantically Rich Local Dataset Generation for Explainable AI in Genomics](https://mlsanity.com/paper/2407.02984/) (2024)
- [RNAGenScape: Property-Guided, Optimized Generation of mRNA Sequences with Manifold Langevin Dynamics](https://mlsanity.com/paper/2510.24736/) (2025)
- [From DNA Design to DNA Slimming: Auditable Agentic Discovery of a Deletion-Only Designer](https://mlsanity.com/paper/2609.40143/) (2026)
- [AutoNovo: Agentic Program Search Discovers Transferable Nucleic Acid Design Algorithms](https://mlsanity.com/paper/2610.16232/) (2026)

## Abstract (arXiv)

Designing DNA and protein sequences with improved function has the potential to greatly accelerate synthetic biology. Machine learning models that accurately predict biological fitness from sequence are becoming a powerful tool for molecular design. Activation maximization offers a simple design strategy for differentiable models: one-hot coded sequences are first approximated by a continuous representation which is then iteratively optimized with respect to the predictor oracle by gradient ascent. While elegant, this method suffers from vanishing gradients and may cause predictor pathologies leading to poor convergence. Here, we build on a previously proposed straight-through approximation method to optimize through discrete sequence samples. By normalizing nucleotide logits across positions and introducing an adaptive entropy variable, we remove bottlenecks arising from overly large or skewed sampling parameters. The resulting algorithm, which we call Fast SeqProp, achieves up to 100-fold faster convergence compared to previous versions of activation maximization and finds improved fitness optima for many applications. We demonstrate Fast SeqProp by designing DNA and protein sequences for six deep learning predictors, including a protein structure predictor.