Smooth $\%$MinMax: A Differentiable Relaxation for Codon Harmonization
Abstract
Codon harmonization aims to adapt the coding sequences for heterologous expression while preserving the native-like patterns of frequent and rare codons that may influence local translation dynamics and co-translational protein folding. However, widely used harmonization metrics, such as MinMax, are defined on discrete codon sequences and are, therefore, not readily compatible with gradient-based neural codon design. Here, we introduce Smooth MinMax, denoted as , a differentiable relaxation of the conventional hard MinMax metric, denoted as . replaces the discrete codon-usage values with probability-weighted synonymous-codon usage values and replaces the hard Max/Min branch with a sigmoid-gated interpolation. This formulation preserves the signed interpretation of , while enabling optimization with respect to the synonymous-codon probabilities and learnable parameters. In human-to-Escherichia coli codon harmonization experiments, closely approximates and supports gradient-based profile matching in synonymous-codon probability space. These results suggest as a practical bridge between profile-based codon harmonization and neural synonymous-sequence design.
Cite
@article{arxiv.2607.03881,
title = {Smooth $\%$MinMax: A Differentiable Relaxation for Codon Harmonization},
author = {Yoonho Jeong and Hyunwoo Choi and Ryan Fernandez Medina Hariri and Eok Kyun Lee and Seung Seo Lee and Insung S. Choi},
journal= {arXiv preprint arXiv:2607.03881},
year = {2026}
}
Comments
17 pages, 2 figures