Joint optimisation of amino acid and coding sequence for de novo designed proteins
Abstract
Abstract De novo protein design now produces backbones whose predicted structures are routinely excellent, yet fewer than half of published designs succeed experimentally, and every design must be expressed from a gene before it can be deployed. A natural protein has a native coding sequence to fall back on; a de novo protein has none, so its gene is built from scratch by a codon optimiser that treats the amino-acid sequence as fixed, while inverse-folding models treat the gene as something that happens afterwards. Here we show that the inverse-folding posterior over designed backbones carries a median of 93 nats of positional entropy per design, with no position on any of 862 backbones fixed, and that this entropy can be spent on properties no codon optimiser can reach. We generalise the codon lattice used in exact mRNA design into an amino-acid-degenerate lattice weighted by unconditional inverse-folding marginals, giving a dynamic program that returns the k best (protein, gene) pairs jointly. Five protein-level liabilities lie outside a codon optimiser's reach by construction; what is not given by construction is that 99.1% of 447 published designs carry at least one removable liability, and that three clear for about a fifth of a nat. The joint lattice improves on the fixed-protein codon dynamic program on 447 of 447 backbones, half a nat captures 89% of the achievable gain, and vendor synthesis-constraint violations more than halve. We release JANUS, which takes a…
Read the paper
Related studies
- Regulation of nucleolus assembly by non-coding RNA polymerase II transcripts
- For a Healthy Histone Code, a Little SUMO in the Tail Keeps the Acetyl Away
- Discovering a binary CTCF code with a little help from BORIS
- Article Commentary: Stabilizing the Code–-Methods to Preserve RNA Prove Their Worth
- Foreword to the Special Issue for the RINTC (The Implicit Seismic Risk of Code-Conforming Structures) Project
- Commentary: Promoting Health and Well-being in Pediatric Primary Care Settings: Using Health and Behavior Codes at Routine Well-child Visits
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.