Papers

Simultaneously engineering multiple protein properties remains a major challenge. Existing machine learning-based pipelines for protein engineering often model properties separately, failing to capture their dependencies and trade-offs. Here, we systematically evaluate how Bayesian parameterization on Multitask Neural Networks can enable robust simultaneous protein engineering under scarce, noisy experimental data. We curated a comprehensive set of 27 multiparameter protein datasets. Then, we compared three algorithm architectures spanning low to full Bayesian parameterization across 16 sequence representations and dimensionality reduction (2,592 models). Bayesian Last Layer models delivered the strongest overall accuracy, generalization, and calibration, ranking as the top-performing model on 70% of benchmark datasets. Dimensionality reduction improved predictive performance by up to 42% and enhanced calibration up to 57% across architectures. Notably, simple One-Hot encoding achieved top performance on 25% of benchmark datasets, particularly with larger datasets. These results establish practical design principles for reliable and data-efficient multiparameter protein engineering.

Single-particle cryo-electron microscopy images a macromolecule as many noisy tomographic projections of its electrostatic potential. We reconstruct the protein backbone directly from such projections, as an atomic point cloud, without the intermediate step of reconstructing the 3D electrostatic potential map. We formulate this as an indirect shape matching problem: a point-cloud template of the backbone is deformed until its simulated projections agree with the data, with the structure observed only through the imaging operator. The deformation is computed via a gradient flow on a Lie group, and we derive the resulting framework in a general geometric setting before adapting it for single-particle cryo-electron microscopy. On synthetic data, we recover single- and multichain proteins and capture conformational transitions.

Biomolecular design underpins applications from molecular recognition to therapeutics and synthetic biology, yet de novo interaction design remains challenging-especially for DNA/RNA, underexplored non-protein modalities with scarce, heterogeneous complex data and sharper geometric and chemical constraints. We introduce MCTH (Monte Carlo Tree Hallucination), an inference-only framework that casts all-atom sequence-structure co-design as uncertainty-aware planning over hallucinated states from pretrained folding and inverse-folding models, with optional biophysical control within the same decision loop. MCTH treats these models as frozen black-box operators and uses Monte Carlo Tree Search to allocate a fixed inference budget across competing design trajectories, incorporating model confidence and uncertainty, as well as cross-expert consensus/disagreement when multiple predictors are available. Across protein-RNA, protein-DNA, protein-protein, and protein-ligand design, matched-budget experiments show that adaptive search improves over simpler sampling and cycling strategies, while held-out AlphaFold3 and Chai-1 evaluations demonstrate transfer beyond the search-time oracle. MCTH provides a shared planning layer across modalities while allowing task-specific folding, inverse-folding, and biophysical modules, requiring no fine-tuning or backpropagation through component models.

We introduce MultiStructRNA, a unified Python toolkit for RNA secondary structure prediction, ensemble analysis, and visualization. Although RNA secondary structure is central to RNA biology and therapeutic design, practical adoption is often hindered by fragmented tooling, incompatible input and output formats, and limited visualization support. MultiStructRNA addresses these challenges through a single high-level API that orchestrates multiple prediction algorithms, harmonizes results into a consistent schema, and provides reproducible, ensemble-aware metrics through an object model suited to both interactive notebooks and production pipelines. MultiStructRNA enables seamless switching between prediction methods without requiring workflow changes and supports both in-notebook and exportable visualizations. Designed for scalability, it supports high-throughput analyses and simplifies comparison across methods while standardizing downstream feature extraction. The current release also includes optional agent-readable workflow recipes that document dependency setup, backend-adapter conventions, SHAPE-data reconciliation, structure interpretation, and comparative sequence analyses. By integrating diverse RNA secondary structure packages within a common framework, MultiStructRNA streamlines structure analysis and facilitates its use in RNA design, optimization, and machine learning workflows.

Local chemical perception and property reasoning are both essential for understanding how molecular structure determines properties. Current LLM-based chemical reasoning methods either receive SMILES/molecular images together with descriptions of local motifs, or reason directly from molecular images. Neither approach enables the model to focus on chemically meaningful regions before reasoning. To address this gap, we propose Visual Latent Structural Reasoning (VLSR), an end-to-end framework that jointly learns localization and reasoning from molecular images. Central to our approach is a localize-then-reason strategy. VLSR first learns to locate chemically meaningful regions in a molecular image. It then reasons about their property effects in a compact latent workspace before producing the final answer. Under the same inference setup, this design achieves 9.6X higher throughput than a comparable textual-reasoning baseline.

Protein language models (PLMs) have transferred the latest advances from natural language processing to computational biology. These models, trained on large corpora of protein sequence data, are widely used to translate amino acid sequences into latent-space embeddings, ready for use in diverse downstream tasks (DTs). By consensus, embeddings from the models' last layers are used, while the models' internal behavior remains poorly understood. We analyzed 13 PLMs across 15 DTs and 9 datasets to assess the value of embeddings from intermediate PLM layers. We trained probe models on embeddings from each layer, compared their performance, and showed that the last layers of PLMs rarely produced embeddings that led to the best results on downstream tasks. Furthermore, we identified a connection between how models learn a certain DT and the similarity between that DT and the pre-training objective. For example, for residue-level downstream tasks, we observed a steady increase in performance across almost all PLM layers, which we attributed to their similarity to most PLMs' pre-training objectives. To allow the community to capitalize on our findings, we provide PLMSommelier, a Python package that automatically identifies the best PLM layer for a given DT with ~98% accuracy and creates a truncated model using only the early layers up to the best-performing layer. This will help users save time and memory during inference and yield better predictive performance.

Protein backbone generation models are often credited with exploring novel fold space based solely on low full-chain similarity to known proteins, yet this cannot distinguish a genuinely new fold from a novel assembly of known structural units. We first ask whether this granularity mismatch alone explains the reported rates, and introduce the Domain Retrieval Rate (DRR), the fraction of generated backbones for which any constituent domain matches a known domain in CATH S40. Applied to eight backbone generation models spanning diffusion and flow-matching paradigms, DRR finds locally alignable known structure in most outputs, while the fraction containing a substantially covered complete domain is considerably smaller and depends on the scoring convention. To calibrate what retrieval alone can achieve, we propose RetFold, a zero-training baseline that constructs backbones by retrieving CATH domains and refining inter-domain connections through geometry-based helix-linker optimization, at two orders of magnitude lower cost on CPU alone.

Proteolysis-targeting chimeras (PROTACs) induce protein degradation by recruiting a target protein to an E3 ubiquitin ligase, making degradation a joint outcome of the degrader molecule and its biological context. Although public databases contain thousands of structured molecule-target-E3 records, degradation measurements are available for only a small fraction of them. Existing supervised approaches therefore leave most recorded chemical-biological relationships unused. We introduce DegradeQuery, a context-aware prediction framework that converts these label-missing records into a pretraining signal. Its counterfactual tuple pretraining objective contrasts recorded tuples with alternatives formed by replacing the target, the E3 ligase, or both, enabling the model to learn contextual associations without assigning activity pseudo-labels. The resulting representation is then fine-tuned to predict degradation from the complete molecule-target-E3 context. On the official PROTAC-8K benchmark, DegradeQuery achieves an area under the receiver operating characteristic curve of 0.9065 and an accuracy of 0.8500, outperforming the compared methods. Controlled analyses further show that the improvement is primarily attributable to tuple-level pretraining, can be recovered using only label-missing records, and remains complementary to protein language model representations. These findings demonstrate that incompletely labeled PROTAC databases contain useful relational supervision and provide a practical route for learning context-aware degradation predictors from scarce experimental labels.

Aquaporins (AQPs) are an ancient group of channels that arose early in phylogeny and radiated throughout the kingdoms of life. Classified as members of the broad Membrane Intrinsic Protein family, AQPs originally were envisioned as strictly water pores that were constitutively open. International research interest is driving new views of AQPs as multi-functional channels with complex control mechanisms that enable adaptive responses to physiological challenges. Expanding numbers of AQP classes are being discovered to allow permeation of diverse solutes in parallel to the archetypal substrate water, with channel activities that show subtype-specific regulation by intracellular and extracellular signals. To date, at least fifteen classes of AQPs from mammals, plants, insects and algae have been proposed to carry ions, enabling amino acid sequence comparisons that suggest ion channel functional domains arise from conserved transmembrane, loop and teminal regions of the tetrameric AQP protein, serving as selectivity filters, gates, and sites for modulation. More AQP ion channel classes likely await discovery, pending identification of relevant activators. Subtype-specific control of dual water-and-ion AQP classes orchestrate diverse roles for AQPs across the domains of life, promoting homeostasis, cell motility, nutrient acquisition, redox protection, sensory detection, and intercellular signaling in cells and tissues.

Proteins occupy heterogeneous free-energy landscapes in which high-entropy ensembles converge toward compact, low-energy basins with multiple sub-states. Molecular dynamics can access these landscapes at atomic resolution, but exhaustive sampling remains computationally demanding. Meanwhile, most quantum approaches target only single optimal structures, leaving full ensemble energetic heterogeneity unexplored. We introduce a residue-level, gate-based quantum circuit framework for coarse-graining protein thermodynamics. Each amino acid is represented as a two-state qubit (stabilised vs. excited solvation state) based on residue solvation energetics. A structure-informed entanglement block then encodes covalent and non-covalent contacts using parameterised controlled gates, embedding correlations across the residue-interaction network. Sampling the circuit ($\sim 10^6$ measurements) yields binary thermodynamic microstates used to compute protein energy distributions, residue-level statistical couplings, energetic sensitivities, and information gains relative to total free energy. We showcase the framework on the benchmark Trp-cage miniprotein 1L2Y (TC5b) and 9GDL, a disulfide-stabilised Trp-cage-fortified exenatide chimera. For 1L2Y, the circuit reproduces a structured, folding-funnel-like energy distribution. Comparative analysis with 9GDL reveals shifts in global energy distributions and residue-level stability profiles. Coupling and information-theoretic analyses localise residues associated with ensemble reorganisation, while multi-body couplings show the circuit resolves both direct and indirect statistical correlations. This framework expands quantum protein modelling beyond single-structure optimisation toward ensemble-level characterisation, capturing key features of rugged energy landscapes to guide protein design, mutation mapping, and allosteric pathway identification.

Oligonucleotide melting temperature is a fundamental determinant of nucleic acid hybridization and underpins the design of molecular diagnostics, polymerase chain reaction assays, and many other biotechnology applications. However, accurately predicting melting behavior remains difficult because it depends not only on sequence composition, but also on experimental conditions and chemical modifications commonly used in modern assay design. Existing thermodynamic models rely on fixed parameterizations that are often difficult to extend across diverse reaction environments and nucleotide chemistries. Here we show that a condition-aware nucleotide language model can accurately predict oligonucleotide melting behavior across diverse experimental conditions and both unmodified and chemically modified oligonucleotides. By combining contextual sequence representations with explicit information describing the reaction environment, the framework achieves sub-degree prediction accuracy and reduces prediction error for locked nucleic acid-modified oligonucleotides by up to 25% relative to nearest-neighbor thermodynamic approaches. The model also more accurately captures the thermal effects introduced by nucleotide modification and maintains strong performance on independent benchmark datasets spanning experimental conditions substantially different from those represented during training. Our results demonstrate that learned sequence representations can complement classical thermodynamic models by capturing context-dependent effects that are difficult to encode using fixed parameter tables alone. More broadly, this work provides a scalable framework for predicting oligonucleotide melting behavior across diverse chemistries and assay conditions, supporting more reliable molecular assay design.

Net charge is a developability-relevant property of therapeutic binders, linked to viscosity, clearance, nonspecific interaction and aggregation, and antibody screens already use charge-related criteria. Yet inverse-folding pipelines expose no way to set it to a target value. ProteinMPNN and BindCraft offer amino-acid biases, weight choices and custom losses, but neither supplies a per-protein feedback loop that measures realised charge after sampling and corrects it to a setpoint. ZetaDial contributes a post-sampling, per-protein secant controller around fixed-backbone ProteinMPNN. On matched stochastic benchmarks the secant loop reduced mean absolute error relative to a fixed-slope loop on RCSB complexes (5.17 vs 6.46 charge units) and Cas13 monomers (5.57 vs 8.23). Relative to the optimised matched global bias, it cut RCSB error from 11.71 to 5.17 (cluster bootstrap p < 0.001) and was statistically indistinguishable on Cas13 (5.47 vs 5.57). Across 800 eight-protein subsets, sensitivity heterogeneity was associated with calibration gain (Pearson r = 0.79); this is descriptive resampling, not a prospective decision rule. Foldability deteriorated as bias magnitude increased. In the full 52-complex seed-0 analysis, reference-based DockQ declined clearly at +/-3 but not at +/-1.5; a selected five-seed replication on eight complexes showed paired declines at every nonzero setting, but does not estimate the effect for all 52. In exploratory BindCraft sweeps, PD-L1 designs moved toward near-neutral charge at similar maximum interface pTM but with overlapping success-rate intervals; IL-7R-alpha responses were non-monotonic and RBD produced no strong designs. A fixed-backbone C-alpha-neighbour analysis found smaller same-sign charge-patch proxies near neutral charge, but this proxy is not a measured electrostatic surface or experimental developability endpoint.

Recent AI advances have enabled protein structure prediction at near-experimental accuracy, largely solving the problem of identifying a dominant conformation from sequence. Many proteins, however, function as dynamic systems populating multiple conformational states with activity emerging from shifts in relative occupancy–an incomplete picture when reduced to one structure. Here, we argue that structure prediction should be reformulated as a state-space inference problem: recovering not one conformation's coordinates but accessible states, their energetic and kinetic relationships, context dependence, and responses to perturbations. We review emerging strategies–deep learning ensemble generators, physics-based simulations, and experimental constraints–and outline a roadmap toward state-space prediction.

A vast class of weak, millimolar-affinity molecular interactions governs cellular function, yet their quantitative characterization has remained largely beyond conventional methods. For over a century, biochemistry has worked within a concentration-based framework where molarity scales with molecular number per volume (N/V), and experiments have usually, often implicitly, changed concentration by moving N while holding V fixed. The weak-interaction measurement bottleneck arises from this paradigm: reading weak binding through bulk concentration requires concentrations beyond practical limits, a framework constraint rather than one of instrumental sensitivity. Here we show that shifting experimental control from N to accessible volume V overcomes this bottleneck and opens previously intractable affinity ranges through nanoscale spatial confinement. Controlling V means controlling what biochemists have called "local concentration" and "proximity effects," recasting these long-ambiguous notions as quantitative variables grounded in first principles. Implemented in DNA nanocavities, the approach showed that geometric arrangement alone can override solution-phase binding hierarchies. The same spatial control quantified a protein-peptide interaction of order 10 mM from femtomoles per well, totalling under a picomole per titration. Even so, a standard plate reader gave a signal-to-noise ratio near 10^3, leaving headroom for still weaker interactions. The affinity-and-geometry readout also enabled rational screening for protein-protein-interaction modulators, identifying compounds that enhance weak associations by reweighting local encounters rather than binding tightly on their own or forming a stable ternary complex. Together, this volume-based paradigm and its implementation provide a general strategy for probing and modulating previously inaccessible biochemical phenomena.

Chemical kinetics has long inferred local molecular behaviour through the flask-and-molarity pairing, where well-mixed concentrations serve as the experimental readout. Yet many biological reactions occur in structured environments. Researchers have long recognized that concentration may not carry the same operational meaning in such environments, but even local concepts such as effective molarity usually translate local effects back into a single value with units of concentration. What has been missing is the complementary path: a bench-compatible way to make local structure an experimental variable, rather than only a correction to molarity. Here we show a chemistry-geometry crossover that the flask-and-molarity interface could not make visible. In the micromolar-or-weaker affinity regime, inhibition can switch sharply out of the familiar concentration-and-affinity mode: chemical binding strength no longer determines the response, and the shape of the target's local space does. A bench-compatible interface made this switch measurable by separating bulk dose from local geometry. This blind spot arose from the hidden premise that macroscopic pooling makes a structured local state readable as a single local concentration. The chemistry-geometry crossover breaks that premise: in a structured target environment, a macroscopic assay can remain sensitive to the probability distribution of local states, so collapsing that distribution to one concentration-valued number removes the geometric control axis from the readout. By preserving that axis in the experiment, the interface bypasses molarity's hidden bottleneck and provides a routine experimental route to remeasure and reinterpret molecular interactions in structured space.