Medicine Henry Quentir Medicine Henry Quentir

TuringQ and Shanghai Jiao Tong University Designed Working Defensins and Xylanases With a Photonic Quantum Training Head: arXiv 2611.08434, 5 November 2026

How TuringQ and Shanghai Jiao Tong University put a photonic quantum head inside protein design training

Quentir Medicine Monitor

Evidence-based insights for quantum medicine.

A team from the Shanghai photonics company TuringQ and Shanghai Jiao Tong University has published a protein design model whose training borrows a photonic quantum circuit as a helper, and has taken the proteins it generated into the laboratory. The preprint, posted to arXiv on 5 November 2026, reports antifungal activity for three designed plant defensins and measured enzyme activity for designed xylanases and short-chain dehydrogenases.

The quantum component sits in one specific place: the training loop. A photonic quantum neural network, or a quantum version of a Kolmogorov-Arnold network, serves as an auxiliary head that predicts how soluble a sequence will be, and its prediction error helps shape the shared representation the design model learns from. Once training finishes, every sequence is generated by an ordinary classical process on ordinary computers.

The author list spans a company and a university physics department. Ke-Xin Tang, Zhen-Qiang Weng, Kai Yan, Yao-Yao Lu, Yu-Ze Zhu, Jun-Jie He, Jie-Cheng Yang, Xiang Zhao and Lin Yang are at TuringQ; Hang Song and Xian-Min Jin are at the School of Physics and Astronomy and the State Key Laboratory of Photonics and Communications at Shanghai Jiao Tong University, with further affiliations at the Chip Hub for Integrated Photonics Xplore in Wuxi, Atomology Co. and Hefei National Laboratory. The work is funded by the National Key R&D Program of China and the national Innovation Program for Quantum Science and Technology.

Protein design asks for an amino-acid sequence that folds the required way and survives contact with water and with a production host. The model the team calls DiffSequence-Q starts from a noisy sequence and repeatedly replaces amino acids until a clean sequence remains, a procedure known as multinomial diffusion. Two conditions ride along at every step: the required secondary structure for each individual residue, written in the nine-state DSSP vocabulary, and a single continuous number for the solubility the designer wants.

Solubility is where the quantum part enters. During joint training, a small auxiliary head reads the shared Transformer representation of a sequence and predicts a solubility score, and the error in that prediction is optimized together with the design objective. In the classical reference version the head is a multilayer perceptron. In the two quantum versions it is a photonic quantum neural network, whose optical modes carry the computation, or a quantum Kolmogorov-Arnold network, which replaces fixed network weights with learnable one-variable functions built from a Fourier series of a chosen order.

On the standard CATH v4.2 test split introduced by Ingraham and colleagues in 2019, the pretrained design model gave the highest secondary-structure recovery of the four models compared when the whole sequence is regenerated, ahead of the strongest baseline by 8.96 percentage points on the strict nine-state measure and 4.77 percentage points on the coarse three-state measure. Its three-state recovery was 60.57 percent with the whole sequence masked, 66.61 percent with the first half masked and 63.11 percent with the last half masked. The authors also report that one of the baselines, the multimodal protein language model DPLM-2, produced structures with the highest mean confidence score in all three settings, so the ranking depends on which quality is measured.

Read More
Medicine Henry Quentir Medicine Henry Quentir

Chalmers, Lund and Jülich Designed Lattice-Protein Sequences on IBM's Torino Quantum Processor: What Physical Review Applied Published on 28 September 2026

Quentir Medicine Monitor

Evidence-based insights for quantum medicine.

Most drug-discovery stories about quantum computers ask a folding question: given a chain of amino acids, which shape will it settle into? Protein designers work the other way round. They start from a shape they want, a pocket that grips a drug or a surface that binds a virus, and search for a sequence of building blocks that will fold into it. A team from Chalmers University of Technology in Gothenburg, Lund University and the Jülich Supercomputing Center has now measured how far one of today's quantum processors gets on the first half of that design problem.

Their paper, "Designing lattice proteins with variational quantum algorithms", appeared in Physical Review Applied (volume 26, article 034065) on 28 September 2026, a year after a first preprint on arXiv in August 2025. Hanna Linn, Lucas Knuthson, Anders Irbäck, Sandipan Mohanty, Laura García-Álvarez and Göran Johansson compare two families of variational quantum algorithms on a task they call protein sequence optimization, and they run the more robust family on IBM's Torino quantum processor, a device built on IBM's Heron r1 chip. With circuit settings carried over from noiseless computer simulations, the hardware found the correct sequence in more than 10 percent of its attempts for 17 of the 22 test problems, including the three largest, chains of 27 to 29 units.

The test proteins are simplified models drawn on a flat grid, and the authors say so plainly. The paper is still a careful read for anyone who funds or buys computational chemistry. In one controlled setting, with every right answer known in advance, it shows where current quantum hardware helps on a design task, where it breaks down, and which engineering choices made the difference.

The test bed is the HP model, one of the oldest simplifications in protein physics. Each protein is a chain of beads on a two-dimensional square grid, and every bead is one of only two kinds. H beads are hydrophobic, the water-avoiding kind of amino acid that real proteins tend to bury in their core. P beads are polar. The energy of a folded chain drops by one unit for every pair of H beads that touch on the grid without being direct neighbors along the chain, so the model rewards a compact hydrophobic core. Real proteins have twenty kinds of amino acid, side chains and three dimensions, so the HP model is a laboratory for methods. Its great advantage is that for chains of up to 30 beads, exhaustive classical enumeration has already listed the exact answers, which lets researchers grade a new algorithm without guessing.

Read More