Papers for

computational chemistry software developers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Agents can design therapeutic antibodies with mixed success

Can agents reliably design therapeutic antibodies?

Abstract: Recent work has demonstrated agents driving protein design workflows, either through orchestration of specialized tools or direct manipulation of sequences. Yet the fidelity with which agents follow protein design instructions and the factors contributing to their effectiveness remain understudied. We examine agentic antibody design under two settings: when explicit molecular-editing instructions are provided and when agents receive only previously-scored sequences and are prompted to propose improvements. We call the first task-guided and the second black-box design. For the task-guided setting, we curate ten expert-validated affinity-improvement tasks and show that LLM judges closely reproduce programmatic evaluations of instruction following. We then study two augmentations: access to antibody-relevant tools and to previously scored design history. Tools improve instruction following (IF) and sequence space exploration while reducing optimization performance, whereas design history reduces IF but improves optimization performance at the expense of exploration. Finally, we combine taskguided and black-box design within an in silico optimization loop and show that our agentic optimizer performs comparably to an established Bayesian optimization baseline.

Sat 31 OctMachine Learning
The gist
Designing better antibodies for therapies is hard and sometimes done by computer programs called agents. This paper studied how well these agents follow instructions when designing antibodies and what helps them do better. The authors tested agents working with clear instructions and agents working by improving existing designs without instructions. They found that tools help agents follow instructions but hurt how well they optimize, while learning from design history has the opposite effect. Combining these methods, agents worked about as well as a standard optimization technique.
Open → 2611.00558v1

Latent jepa helps predict chemical outcomes from hidden reasoning steps

Latent JEPA: Abstract Future Prediction for Latent Reasoning in Chemistry

Abstract: Large language models offer a promising foundation for chemical reasoning, bringing together chemical knowledge and multistep problem solving. Chemical intuition can provide an initial sense of plausible outcomes before the details of a solution are fully worked out. Inspired by how such expectations complement explicit analysis, we study how continuous latent thoughts can be trained to anticipate informative aspects of future solutions without verbalizing every intermediate step. We introduce Latent JEPA, a framework that combines autoregressive learning with joint-embedding prediction of one or more future views. For chemical reasoning, we develop textual and molecular prediction objectives that connect latent thoughts to both subsequent reasoning and molecular outcomes. Experiments on ChemCoTBench show gains in molecular optimization and on several editing and reaction metrics. Representation analyses show that future prediction makes latent thoughts more informative about molecular outcomes and strengthens their correspondence with chemical structure. These findings support abstract future prediction as a learning principle for connecting continuous latent reasoning with scientific outcomes.

Thu 1 OctMachine LearningComputation and Language
The gist
Predicting chemical reactions and molecular changes is tricky because it involves many steps and possibilities. The authors made a new method called Latent JEPA that trains a computer model to anticipate important future results without explaining every tiny step out loud. This helps the model think ahead and connect what it 'thinks' internally to actual chemical outcomes better. When tested, this approach improved the model’s ability to suggest useful changes to molecules and predict reactions.
Open → 2610.01947v1

Disordered crystal prediction model works without detailed site labels

EP-Flow: Disordered Crystal Structure Prediction without Site-Level Annotations

Abstract: Generative models have made rapid progress in ordered crystal structure prediction, yet many functional materials are intrinsically disordered, with substitutional mixing, vacancies, or interstitial species controlling their properties. Existing crystal generators either assume deterministic site occupations or require site-level disorder annotations, which are often unavailable when the chemical formula is the primary input. We formulate disordered crystal structure prediction through an Occupancy Distribution Matrix (ODM), a continuous site-by-species representation that unifies ordered crystals, solid solutions, vacancy disorder, and interstitial occupancy. A valid ODM must satisfy coupled site-wise occupancy, mass-conservation, and non-negativity constraints, placing each sample on a formula-dependent transportation polytope. We propose Entropic Polytope Flow (EP-Flow), a marginal-constrained flow matching framework that canonicalizes heterogeneous polytopes into a shared double-centered space, learns a marginal-preserving flow, and recovers feasible occupancies through a Sinkhorn inverse map. By jointly generating occupancies, fractional coordinates, and lattice parameters, EP-Flow achieves state-of-the-art performance on formula-conditioned disordered CSP benchmarks derived from COD and MPDS, substantially outperforming adapted ordered-crystal generators. Analyses further show that EP-Flow recovers sparse and chemically meaningful local disorder patterns rather than merely matching global composition statistics.

Thu 1 OctMachine Learning
The gist
Predicting the atomic structure of materials that have mixed or missing atoms is hard because usually detailed information about each atomic position is needed. The authors developed a new mathematical way to represent these messy materials using a continuous matrix called an occupancy distribution matrix (ODM). Their new method, EP-Flow, uses advanced flow-based modeling and special constraints to generate realistic crystal structures from just the overall chemical formula. This approach works better than previous models that needed more detailed inputs and can find meaningful patterns of disorder in materials.
Open → 2610.01315v1

TopU-LBVS provides realistic multi-target drug screening benchmark

TopU-LBVS: A Realistic Multi Target Benchmark for Ligand Based Virtual Screening

Abstract: Ligand-based virtual screening (LBVS) is a practical first-pass tool in early-stage drug discovery, but existing benchmarks can overestimate performance through random negatives, easy decoys, limited target coverage, and non-standardized evaluation protocols. We introduce TopU-LBVS, a multi-target benchmark for LBVS under hard-negative screening conditions. Starting from curated ChEMBL~35 bioactivity data, TopU-LBVS covers 93 protein targets across 7 protein classes and constructs target-specific screening libraries with property-matched, structurally similar decoys at a fixed 1:40 active-to-decoy ratio. Libraries contain roughly 400 to 10,000 compounds and are designed to reduce simple physicochemical and nearest-neighbor fingerprint shortcuts. TopU-LBVS provides three fixed protocols. TopU-LBVS-full evaluates ChEMBL$^\ast \rightarrow$ TopU generalization across all 93 targets. TopU-LBVS-low evaluates low-data TopU $\rightarrow$ TopU learning within the hard-negative distribution. TopU-LBVS-mini provides a compact seven-target protocol with a paired random-decoy control that changes only the test decoys, enabling low-cost development and direct measurement of the gap between random ChEMBL$^\ast$ and TopU decoys. Across ten reference baselines spanning fingerprint methods, molecular GNNs, fingerprint hybrids, and modern molecular models, performance under random-decoy evaluation degrades sharply under hard-negative screening. We release data, fixed splits, evaluation code, and baseline implementations for reproducible comparison of future LBVS and molecular representation learning methods. Code and data are available at https://github.com/topu-benchmark/topu-lbvs and https://huggingface.co/datasets/topu-benchmark/topu-lbvs.

Thu 24 SeptMachine LearningArtificial Intelligence
The gist
Early drug discovery often uses virtual screening to find molecules that might interact with proteins, but many existing tests make it look easier than it really is. The authors created TopU-LBVS, a challenging test set with many protein targets and realistic 'hard-negative' molecules that resemble actual drugs but are inactive. They show that popular methods perform worse on this tougher test, suggesting it better reflects real-world difficulties. The benchmark includes clear protocols, data, and code to help developers fairly compare new virtual screening methods.
Open → 2609.29740v1

Complete neural initialization speeds up materials density calculations

Complete Neural Electronic Initialization Accelerates Materials DFT

Abstract: We present the first complete machine learning method for accelerating plane-wave density functional theory (DFT) in materials under the projector augmented wave (PAW) formalism. We formalize seven criteria that a \textit{Complete Neural Electronic Initializer} must satisfy for practical end-to-end PAW DFT acceleration. Applying these criteria to prior work reveals two missing structure-dependent components, augmentation occupancies and spin initialization, that prevent existing methods from providing complete reference-free initialization. Controlled ablations show that omitting these components can eliminate or reverse the acceleration obtained via models that only predict the smooth valence density. We satisfy these missing requirements by introducing AugNet, the first general equivariant model for PAW augmentation occupancies, and the first general spin density model for materials, which predicts the smooth spin-difference density and spin-difference PAW augmentation occupancies using predicted magnetic moments to constrain the global magnetic state. Combined with existing valence density models, these components satisfy all seven criteria and form a fully reference-free electronic initializer for materials DFT, requiring no electronic quantities from a converged target calculation. Our method reduces end-to-end DFT wall time by up to ~25% on unseen structures while preserving converged energies.

Fri 18 SeptMachine Learning
The gist
Calculating the properties of materials at the atomic level is very slow because it needs to figure out how electrons behave. The authors found that previous AI methods for speeding this up missed important details. They created new models that include these missing parts and can start the calculations from scratch without extra information. This new approach speeds up the calculations by about a quarter while keeping results accurate.
Open → 2609.21759v1