bioinformatics developers

Abstract: Unsigned linear unichromosomal genomes with genes labeled 1, . . . , n are represented by permutations of these labels, whose adjacencies are the unordered pairs of genes occurring consecutively. The breakpoint distance counts the adjacencies of one genome that are absent from the other, and a breakpoint median minimizes the total breakpoint distance to a given set of genomes. Deciding whether a permutation is a median often requires a minimization over the whole symmetric group, which makes the median set resistant to exact counting. We replace this global condition by a local one: a c-approximate median is a genome having at most c adjacencies absent from all the input genomes. For a fixed genome we introduce its approximate median inverse, the set of ordered k-tuples of genomes ad- mitting it as a c-approximate median, and we construct and count this set explicitly. For k genomes chosen uniformly and independently at random, this gives the probability that a given genome is a c-approximate median and the expected number of such approximate medians. The relaxation is not only technical: approximate medians and near-optimal genomes co- incide up to a shift of the tolerance that remains bounded in probability as the number of genes grows. Finally, we derive an explicit upper bound on the probability that a given genome is an exact median, and on the expected number of exact breakpoint medians.

Abstract: The genetic code exhibits redundancy, meaning that multiple distinct nucleotide sequences can encode the same amino acid sequence, a property mathematically characterized as non-injectivity. While this phenomenon is well studied in molecular biology, its rigorous verification within a computational framework remains limited. In this work, we present a machine-verified model of the central dogma of molecular biology using the Lean 4 interactive theorem prover. We formalize nucleic acid bases and amino acids as inductive types and define transcription and translation as functions mapping DNA sequences to RNA and subsequently to polypeptides. Within this framework, we prove key properties of these biological processes, including the redundancy of the genetic code, injectivity of transcription, and the non-injectivity of translation. Additionally, we formalize auxiliary processes such as DNA replication and mutation, and verify properties such as length conservation under transcription and point mutations. Finally, we demonstrate how Lean 4 helps prevent semantic errors in our code, showing its effectiveness as a tool for advancing reliable, bug-free scientific computing, particularly within the life sciences.

Abstract: Public drug-protein bioactivity resources commonly identify targets by UniProt accession, although one accession may represent distinct experimental sequences. We present Curate-DTI, a construct-aware dataset derived from the 1 July 2026 BindingDB export, with IC50, Ki, Kd and EC50 processed independently. The release contains 2,919,717 measurements, 1,291,872 standardized ligands, 9,668 sequence-defined protein entities and 2,275,683 endpoint-specific interactions spanning 8,336 reviewed UniProt accessions. Protein entities combine the accession with the reported experimental sequence, preserving constructs that accession-only processing would merge. Across the four endpoints, 4,204 endpoint-specific ligand-accession groups showed different classification labels across sequence-defined protein entities. Exact and inequality-qualified observations provide relation-aware classification evidence, whereas regression targets use exact values only. Connected Parquet tables preserve individual measurements, standardized structures, protein sequences, aggregated interactions and available assay and source provenance. The resource supports endpoint-specific classification and regression while allowing users to select molecular representations, protein representations and evaluation splits. Preprocessing code, a column dictionary, exclusion reports and validation summaries accompany the release.

Abstract: Multi-stage analytical tasks require LLM agents to use intermediate evidence to decide whether to continue, redirect or stop. Self-reflection, in which an agent critiques its output, is widely studied, yet a critique does not show whether evidence changed the next action. We call that capability self-restraint. Existing agent benchmarks largely measure task completion or output quality and leave it untested. DMAIC-Bench evaluates it with 97 seeded tasks across 15 industrial cases in four sectors, built around Lean Six Sigma's explicit statistical prerequisites. Paired counterfactual variants require opposite actions at the same workflow position, distinguishing warranted restraint from indiscriminate stopping. We assess execution traces with a private scorer and final claims through independent expert review. Across 13 models from six providers, execution-level routing accuracy averages 92% on matched GO decisions and 19% on NO-GO decisions, counting any forbidden downstream computation as a route error. Execution traces show agents reporting correct diagnostics before performing analyses that those results disallow. Separately, blinded expert review of the 11 audited agents estimates that reports act on invalid evidence at 47% of reached NO-GO decisions, with an interval from 31% to 61%. Explicit stopping permission helps some models without consistently restoring correct decisions at both GO and NO-GO forks. Sixteen life-sciences tasks in bulk ribonucleic acid sequencing (RNA-seq) and case-control genome-wide association studies (GWAS) show that the same framework can evaluate evidence-dependent decisions beyond industrial workflows. Project code and data are available at: https://github.com/psptpx/DMAIC-Bench-public.

Abstract: Barycenter models, based on stochastic and deterministic optimal transport bridges between probability measures, have proven to be useful for supervised and unsupervised learning for (probability) measure-valued data. In this paper, we utilize the directional asymmetry and dual drift-diffusion control arising in the recently proposed Schrödinger-Bass Bridge (SBB) between two measures, and propose \textit{SBB-encoder}, \textit{SBB-decoder}, \textit{SBB-symmetric}, and \textit{SBB-Markovian} barycenter models. For each of these novel models, our first theoretical result establishes the existence of the SBB-barycenter under mild assumptions. Our second theoretical result characterizes the decoder SBB-barycenter of Gaussians where we show that for Gaussian input measures, the decoder formulation admits a Gaussian minimizer, reducing barycenter computation to a finite-dimensional optimization and yielding a compact algorithm. For general input measures, we derive a hybrid Wasserstein gradient particle descent scheme for all four formulations using the SBB dual representation. Finally, we utilize these algorithms for solving distributional-inverse/interpolation problems arising in the context of spatial images and single-cell RNA sequencing data, where we show that the SBB-decoder and Markovian barycenter allow us to interpolate between successive time points. Results show that our method outperforms existing barycenter models for measure interpolation and inference.

Abstract: Accessible research software can connect published antimicrobial peptide (AMP) records to sequence analysis, molecular visualization, and interactive design without requiring a dedicated application server. We present AMP Atlas, an open-source, browser-only workbench that combines a DRAMP-derived database with local physicochemical calculations, published-neighbor retrieval, and sequence-guided structural comparison. The reported snapshot contains 12,784 source records, 10,579 canonical records, 5,183 source SMILES annotations, and 433 PDB chain structures. An inverse helical-wheel editor lets users specify peptide length, fill residues gradually, and immediately retrieve exact-length database matches under positional and property constraints. Selected-position residue frequencies support further editing while preserving the remaining constraints. A Web Worker performs searches and numerical calculations; a WebGL viewer provides interactive molecular representations. The static distribution occupies 138.8 MB and supports GitHub Pages hosting. Verification includes 23 Python tests, 30 browser-core/worker checks, a designer DOM integration suite, and agreement against fixed numerical references. This report describes the implemented algorithms and assumptions; it does not establish predictive accuracy for antimicrobial activity or experimentally validate designed peptides. The website is available at https://hysonlab.github.io/amp-atlas/ and the source code and database are available at https://github.com/HySonLab/amp-atlas

Abstract: Large protein language models provide a powerful framework for learning sequence-function relationships and guiding experimental design. However, architectural improvements are underexplored, especially in low-compute regimes of relevance for resource-constrained research laboratories. We show that the training cost for encoder-only transformer architectures can be reduced by leveraging modern architectural refinements, offering an encouraging performance / compute-efficiency trade-off. We introduce ProtModernBERT, a modernized BERT-style model pre-trained on amino acid sequences, and compare it with ProtBert, Amplify, and ESM2 across a set of supervised structural tasks -encompassing secondary structure prediction, disorder prediction, and subcellular localization- and ProteinGym's zero-shot functional benchmark. We find that ProtModernBERT matches or closely tracks the performance of Amplify-120M and ProtBert with up to 80-90\% reduction in pre-training FLOPs, and narrows the gap to ESM2 when controlling for pre-training. Our analysis using a cluster-based validation protocol for hyperparameter selection and randomly initialized models disentangles the contributions of architecture and data, showing that modern components provide stronger inductive biases that are more consequential than parameter count in low-compute regimes.

Abstract: DNA methylation is a critical epigenetic modification that regulates gene expression and genome stability, primarily catalyzed by the DNA methyltransferase 3 (DNMT3) family of enzymes. Distinguishing precise sequence-specific flanking preferences between DNMT3A and DNMT3B remains computationally challenging due to the complex and high-dimensional nature of biochemical sequence data. In this study, we propose a novel Quantum Machine Learning (QML) approach to classify DNA methylation preferences, leveraging a Variational Quantum Classifier (VQC). Using experimentally derived deep enzymology data, we constructed a dataset containing 4,096 unique NNNCGNNN motifs. We evaluated a four-qubit VQC using a custom NumPy exact-statevector simulator. The described two-layer circuit contains 24 trainable rotations, eight CNOT gates, and eight input-encoding rotations, totaling 40 gates before measurement. Across five initialization seeds, the revised VQC evaluation achieved a ROC-AUC of 0.661 pm 0.009, where variability denotes sample standard deviation. The held-out test partition is approximately balanced, containing 200 DNMT3A-labeled motifs and 210 DNMT3B-labeled motifs. An always-DNMT3B predictor achieves 51.2% accuracy and a balanced accuracy of 0.500 on this test partition. This proof-of-concept establishes a foundation for utilizing near-term quantum architectures in complex epigenetic classification tasks, paving the way for hybrid quantum-classical pipelines in bioinformatics.

Abstract: This paper presents a data-driven approach to lossless and lossy source compression using polar codes. Classical polar source coding achieves fundamental limits with low complexity for binary memoryless sources. However, for unknown q-ary sources with memory, successive-cancellation (SC) decoding requires an explicit source model and can be computationally prohibitive. We present analytical SC recursions and show that exact implementations scale quadratically with alphabet size and cubically with state-space size. To overcome these limitations, we develop neural polar decoders trained directly from data to approximate conditional distributions of synthetic polar coordinates without requiring an explicit source model. These learned distributions enable estimation of entropy rates, mutual information, and rate-distortion functions, and support explicit polar compression schemes. For lossless compression, we identify high-entropy polarized coordinates and reconstruct the remaining coordinates using the learned SC recursion. For lossy compression, we combine neural polar decoders with a neural test channel to estimate and optimize the mutual information-distortion tradeoff and construct an explicit lossy polar code. We also extend the framework to Wyner-Ziv compression with decoder side information by learning conditional polarized distributions and a reconstruction mapping. Experiments on memoryless sources, sources with memory, and real genomic data show that the learned decoders closely match analytical SC decoders when available and remain applicable when conventional methods are computationally infeasible or the source model is unavailable.

Abstract: Processing-in-Memory (PIM) on monolithic 3D (M3D) DRAM is a promising answer to the memory wall for data-intensive dynamic programming (DP), yet extracting its performance today demands hand-written kernels: the programmer must pick a tile size, place data across non-uniform-latency memory tiers, partition work between heterogeneous processing units, and insert the right broadcasts and queues by hand. We present GenMLIR, an MLIR compiler that automates this mapping for tier-aware 3D PIM. GenMLIR encodes the semi-ring generalized grid update, the algebraic form shared by all-pairs shortest path (APSP) and sequence alignment, as a first-class IR abstraction, and lowers it through four GenDRAM-aware pass groups that derive blocked tiling, tier-aware placement, tile-to-PU assignment, and explicit communication. We also characterize precisely which DP recurrences the abstraction admits and which it does not. On the GenDRAM architecture, GenMLIR-generated code runs up to 5.8 (APSP) and 16.7 (alignment) faster than a lowering without compiler support, and 1.2-1.4 faster than a standard affine-tiling PIM compiler we also implement, reaching 90-100% of an achievable-performance bound. It expresses these workloads in 4-9 lines instead of 300-500 and compiles in a negligible fraction of runtime.

Abstract: AlphaFold 3 predicts protein structures with remarkable accuracy, yet how structural information emerges within the model remains poorly understood. Here, through causal interventions on internal representations and direct probing of every Pairformer block, we trace the formation of global protein geometry and identify the multiple sequence alignment (MSA) as a structural shortcut to the fold. Removing the MSA largely preserves local secondary structure while disrupting the long-range relationships that define global topology. Restoring the MSA-enriched pair representation at only forty residues recovers most of this lost organization, including at pairs never directly modified. This contribution depends on the detailed direction of the MSA module's output rather than its magnitude. The Pairformer rapidly converts this signal into global geometry: the final fold becomes recoverable by approximately block 9 of 48 for a majority of proteins, roughly twenty-seven blocks before the model's decoder can render it, whereas without the MSA it remains inaccessible for most proteins throughout the pass. Which homologs are supplied shapes this trajectory more strongly than which query is supplied; it persists for a designed query that never evolved but collapses for a shuffled sequence. Most importantly, an alignment built for a different protein that shares the fold, supplied only at the structurally corresponding columns, raises the median TM-score against experiment from 0.44 to 0.72, while the same alignment shifted a few residues along the chain performs worse than supplying no alignment at all. What AlphaFold 3 reads from an alignment is therefore a description of the fold itself, transferable between proteins that share one, rather than the query's own evolutionary history. This explains both its accuracy and the limits of what it has solved.

Abstract: Identifying upstream regulators that connect several disease processes to therapeutic interventions is a central objective in Alzheimer's disease drug discovery. We propose an entropy-guided reverse-causal framework that makes candidate bottleneck genes the organizing link between disease mechanisms, pathways, molecular targets and drugs. The methodology integrates five stages: an Alzheimer's-specific knowledge graph with language-model assistance and expert review; reverse tracing from drugs to candidate genes; entropy-guided prioritization; forward propagation to drugs and complementary combinations; and staged validation with evidence feedback. The novelty lies in integrating upstream bottleneck identification, entropy-guided prioritization and iterative therapeutic selection within a dynamic, bidirectional discovery architecture. We demonstrate its molecular tracing and gene-prioritization components in a computational feasibility study using DeepDrug2 and MSigDB pathway annotations. Tracing amlodipine, indapamide and atorvastatin through a network of 11,300 molecular and drug nodes identifies 46 routes to nine genes. EGFR is the leading candidate, supported by 26 routes from all three drugs; MME and MAF rank next. These results show how pharmacological starting points can identify shared candidate genes with defined molecular connections. The framework's scientific significance lies in connecting convergent disease mechanisms to systematic intervention selection, with preservation of cognition and independence as the translational objective.

Abstract: Ribonucleic acid (RNA) functions through conformational changes that are not fully captured by static structures. However, large-scale standardized RNA dynamics data remain limited, and existing approaches typically treat trajectory generation and dynamics understanding as separate objectives. Here, we introduce RNADynBench, a standardized RNA molecular dynamics (MD) benchmark with 2585 quality-controlled 100-ns all-atom trajectories and leakage-controlled splits. Building on RNADynBench, we develop RNADynNet, a unified model for RNA dynamics learning that uses a shared backbone for both trajectory generation and dynamics fingerprint extraction from a single conformer. It combines coordinate denoising, single-frame-to-trajectory alignment, and physical grounding to connect all-atom trajectory generation with dynamics representation learning. Physical grounding improves both generated dynamics and the physical information recoverable from these fingerprints. Across both test sets, including the high-flexibility challenge set, the generated trajectories achieve RMSF correlations of 0.875 and 0.766, while single-conformer predictions show comparable agreement with MD-derived dynamics. RNADynBench and RNADynNet together establish a benchmark and unified modeling framework for generating and understanding RNA dynamics.

Abstract: Every protein has a unique stability landscape, but the physical consequences of mutation are governed by recurring biochemical constraints. We test whether a shared decoder, trained on measurements from diverse proteins, can interpret these constraints in an unseen target, enabling cross-protein transfer for initial experimental round prescreening. We present StabilityArc , which maps frozen ESMC-600M residue representations through a shared RoPE transformer to an Lx20 matrix of substitution effects; a symmetric, contact-aware residual aids in predicting epistasis in simultaneous substitutions. In 66 strict leave-one-protein-out evaluations covering 134,794 ProteinGym variants, StabilityArc achieves 0.7134 Spearman correlation, exceeding the strongest zero-shot baseline, ProSST-2048 (0.6526), by 0.0608. We further explore the utility of this method by providing the score as a prior for Kermut, achieving Spearman correlation of 0.8280 across three supervised split schemes, improving on Kermut's reported 0.8167.

Abstract: FM-indexes and compressed suffix arrays (CSAs) are often treated as interchangeable, but they behave differently as the alphabet grows. An FM-index step costs about one cache miss per level of a wavelet tree, so it gets slower with the alphabet size. A CSA step is a binary search whose range shrinks as characters get rarer. We describe a simple hybrid. Each character of the text is replaced by the rank of its frequency among the characters that follow the previous character. We backward-search on this encoding, which is over a small, skewed alphabet, and recover the one piece of information the encoding loses (the first character of the pattern) with a single CSA-like step on an array we call $\PsiE$. Counting is exact, and locating works with standard suffix-array sampling. A prototype on synthetic repetitive data shows that the hybrid is the fastest of the indexes we tried at intermediate alphabet sizes with 1\% noise, but even its compact version is 1.7 to 3.9 times larger than a compressed run-length CSA or FM-index of the original text, because the encoding and $\PsiE$ together have more runs than the original Burrows--Wheeler transform. Whether that changes on real data, such as parses and minimizer digests, is the main open question.

Abstract: The Consensus technique marked a breakthrough in the construction of minimal perfect hash functions (MPHFs), reaching a linear tradeoff between construction time and space overhead relative to the optimum. Consensus provides a clever scheme to search for and encode seeds of tasks in random data structures. We apply Consensus to the related field of compressed static functions (CSFs). These data structures store a function $f: S \to Σ$ such that querying a key $x \in S$ returns $f(x)$ and querying $x \not \in S$ returns an arbitrary value. CSFs do not need to store the keys $S$ and only need space close to the zeroth-order empirical entropy of the multiset of values. Often, some values are much more common than others. In these cases, CSFs can use less space than their non-compressed counterparts. CSFs are a useful building block, for example in database design and bioinformatics. We introduce Consensus-CSF, which can reach arbitrarily close to the empirical entropy $n H_0$, with a construction time of $n \exp(\tilde{\cal{O}} (\sqrt{1 / δ}))$ for space usage of $n H_0 (1 + δ)$ when assuming some parameters of the value distribution to be constants. This tradeoff beats previously implemented approaches that can only reach some fixed threshold above the entropy lower bound. We enable Consensus in the setting of CSFs, which is less structured than MPHFs, with the introduction of task insertions. Our approach randomly distributes the keys into one-bit Consensus tasks and then strategically inserts additional tasks in places where the construction would get stuck otherwise. We provide an implemented version of our algorithm which reaches the same order of magnitude in space overhead as competitors but is not competitive in practice. Beyond these results, we present a new way to think and reason about Consensus, which may also be applied to other problems.

Abstract: We spent five weeks using an LLM coding agent on open problems in coding theory: finding large sets of four-letter words, such as DNA barcodes, that stay far apart in edit distance. The agent wrote the verifiers and search code; a human chose the problem and set the verification protocol. Restricting the search to codes with a prescribed symmetry, a classical technique, shrank the problem about fourfold and raised the best known code of length 6 and minimum edit distance 3 from 114 to 120 words ($E_4(6,3) \geq 120$). The same pipeline improved twelve further lower bounds at lengths 6 to 9 and distances 3 to 6. We give the failures equal space. Our own search stopped at 116 and recorded the last symmetry class as topping out at 112; a second agent session, running the same search with a better operator, found the 120. A later verdict that the method did not carry over to length 7 was wrong for the same reason, and an earlier instance cost three weeks. Each time, an intermediate result was written down, never rechecked, and treated as a fact that ruled out further search. Checking final outputs, as our protocol required, does not catch such errors.

Abstract: MeanFlow enables efficient few-step generation by predicting interval-average velocities, but this representation creates a mismatch for reward fine-tuning: existing advantage-based objectives are typically defined on instantaneous velocities or equivalent $x_0$-space predictions, whereas inference directly uses the learned average-velocity map. We introduce MeanFlowAdvantage, a signed advantage-weighted least-squares objective for average-velocity generators. Our key construction uses a shared, detached MeanFlow derivative correction to express the reward objective in prediction space while making rollout and reference regularization exact penalties on the average-velocity network deployed at inference. The resulting formulation preserves MeanFlow's native few-step sampler and provides a direct mechanism for transferring reward improvements to the deployed flow map. On SD3.5-Medium, MeanFlowAdvantage improves all eight reported metrics over the matched four-step MeanFlowNFT baseline and, with only four NFEs, matches or exceeds the 40-step DiffusionNFT baseline on six of eight metrics. The same objective also transfers to DNA promoter design, where it supports both teacher-free on-policy RL for a generator defined on a manifold and teacher-guided reward-graded distillation, with the latter yielding the lowest one-step Sei profile MSE among the compared configurations.

Abstract: RNA design aims to identify sequences that fold into specified secondary structures. Existing methods formulate the task as target-specific search or conditional generation. However, natural RNA evolution proceeds through sequence variation and selection, with compensatory substitutions, whereas these methods do not explicitly model this process. To address this limitation, we propose a two-stage framework comprising RNA Inverse-Folding Flow (RNA-IFlow) and RNA-IFlow-RL. RNA-IFlow uses structure-conditioned Dirichlet Flow Matching to model coordinated variation across the sequence, while RNA-IFlow-RL maps the learned flow to a pairing-preserving finite policy and refines it with thermodynamic feedback. Our framework achieves leading performance on multiple benchmarks, reaching 85.19% Pass@1 on Rfam-27. Further analyses reveal thermodynamic gains, policy dynamics, and robustness across settings. Our work couples coordinated variation with thermodynamic selection, offering a novel paradigm for RNA design.

Abstract: Input encodings can restrict which measured contrasts a predictor can jointly reproduce, even when no single contrast is forced to vanish. We compute the attainable contrast space from an encoder's equivalence classes and a fixed contrast design, without labels, loss, or a fitted model; projecting the recorded contrasts onto that space gives an empirical error floor for any unrestricted decoder on those classes. On a 140-rectangle siRNA interaction panel, a graph neural network's training-only feature mask merges 165 endpoint states into 90 classes and cuts the rank of the 140 interaction contrasts to 72. The resulting floor is 0.009980, which is 14.6% of the fitted model's interaction squared error; the fitted model reaches 0.068335, slightly worse than a control predicting no interaction at all. A minimum of three restored chemistry columns recovers full rank. Refitting without the mask removes the floor entirely, yet interaction MSE improves by only 0.000017 under the reported protocol, and the restored columns remain absent from every training input. On a released RNA-splicing predictor, whose encoding is injective on the measured states, the same computation returns the full design rank of 1,986 and a floor of exactly zero. These results separate what an encoding permits from what a fitted model achieves; they do not identify what limits the remaining error. The rank check needs no fits and bounds what any amount of training under a fixed encoding can recover. The project repository is available at https://github.com/shadi97kh/REPRESENTABLE-BUT-UNLEARNED.

Abstract: Large visual foundation models have demonstrated remarkable transferability across a wide range of downstream tasks. To deploy such models efficiently, feature matching has become a popular knowledge distillation approach that transfers teacher representations to smaller student models without requiring labeled data. However, we show that the conventional feature matching objective with L2-distance is inherently biased toward reconstructing dominant spectral directions of the teacher representation, while under-optimizing low-variance directions that often contain task-relevant information. To address this, we propose Spectrum-Balanced Feature Matching, SpecMatch, a simple objective that adaptively emphasizes under-optimized spectral directions while preserving the relative importance of dominant directions. SpecMatch is easy to implement and introduces negligible computational overhead. Extensive experiments on image recognition demonstrate that SpecMatch consistently improves downstream adaptation across diverse tasks, including image classification, anomaly detection, medical image analysis, and domain generalization. In particular, SpecMatch outperforms conventional feature matching in 40 of 42 teacher--student and training-setting combinations, while consistently improving over the original student model in all settings. We further demonstrate that the proposed objective generalizes beyond vision, improving downstream performance across six protein understanding tasks.

Abstract: Predicting spatial gene expression from histology images could scale spatial transcriptomics (ST) to image-only cohorts, but conventional histology-based ST prediction is trained and evaluated mainly by per-gene spatial-profile reconstruction. This objective is misaligned with a key downstream use of ST: differentially expressed gene (DEG) discovery, where genes are ranked for a biological or morphology-defined contrast by evidence of between-group expression differences. We formulate image-based differential expression ranking (IDER), which asks whether predicted expression profiles preserve the contrast-specific ranked gene list obtained from measured profiles. IDER compares gene rankings induced by differential-expression statistics, rather than raw expression magnitudes or per-gene spatial correlations. We further introduce a differentiable IDER objective that aligns these statistics across genes and can be trained with morphology-derived proxy contrasts without predefined biological group labels. Experiments on public ST datasets show improved DEG-ranking agreement and pathway-enrichment overlap over conventional reconstruction objectives, including morphology-derived and pathologist-annotated tissue-region evaluations.

Abstract: As the bridge between protein modality and discrete modeling, protein structure tokenization still largely relies on heavily engineered training objectives tailored to specific downstream tasks and large training datasets, which hinders its transfer to broader application scenarios. To address this issue, we propose ProFiT, a lightweight flow matching tokenizer. With simple training strategies that encourage healthy codebook utilization, ProFiT can be trained efficiently and naturally learns semantically meaningful representations without any manual semantic alignment, while achieving reconstruction quality and generalization that match or surpass those of substantially larger tokenizers. We conduct extensive evaluations across a wide range of settings and demonstrate that ProFiT is a plug-and-play tokenizer adaptable to diverse downstream tasks. This study further reveals the significant potential of the flow matching tokenizer paradigm. Our code is publicly available at https://github.com/QDKStorm/ProFiT.

Abstract: Rotary position encodings make attention sensitive to relative position, but extending them to graphs requires choosing how graph structure enters the rotation. Previous works assign each node a rotation from spectral coordinates, so the rotary factor between two nodes depends only on their endpoints and cannot distinguish the routes connecting them. We introduce \textit{Analytic-Walk Rotary Positional Encodings} (AW-RoPE), which place the rotations on edges and sum the transported features over all walks, so contributions along different routes can reinforce or cancel. An exact variant evaluates the complete sum by a differentiable linear solve, and a sparse variant truncates it at a finite depth. We prove forward and parameter-derivative truncation bounds at fixed inputs and parameters. Both variants act on projected queries and keys, and the sparse recurrence also augments message-passing networks. Across five synthetic tasks both variants reduce nRMSE by $15$--$58\%$ relative to the strongest baseline, and on real superpixel, peptide and OGB benchmarks the sparse recurrence attains the best mean on every dataset with Performer kernels and on twelve of thirteen datasets with GIN. Analysis shows that AW-RoPE can distinguish routes whose only cue is how two endpoints are connected, while node-wise rotary encodings cannot.

Abstract: AI co-scientists could accelerate computational research, but over a long-running study the workflow also has to stay inspectable, resumable and reproducible, which requires persistent computational state and provenance. Here we present OpenAI4S, an open-source scientific research agent built around the principle of \emph{Code as Action, Science as Sessions}. OpenAI4S combines a persistent computing runtime with research-session management: orchestration is handled through structured tool calls, while scientific actions are represented as complete code cells executed in persistent Python and R kernels. An append-only Action Ledger, per-cell execution records, versioned artifacts, environment records, and workspace checkpoints preserve how results were produced and support session recovery, branching, and extension. Configurable sandboxing, permission controls, and code and trajectory screening provide complementary safeguards. We evaluate OpenAI4S on 36 research scenarios spanning retrosynthesis, molecular dynamics, protein binder design, protein mutation, catalyst screening, and mineral spectroscopy, measuring scientific task accuracy, workflow completeness, and reproducibility of the resulting repositories. OpenAI4S achieves an overall score of 7.83, compared with 5.7--6.4 for a general-purpose coding harness evaluated with three frontier models, with the largest gains on long-horizon and computation-intensive workflows. These results suggest that integrating persistent execution with session-level provenance can improve the reliability of AI-assisted scientific workflows. Environment specification and full rerunnability remain weak for every evaluated system, ours included, so reproducibility is still an open problem for scientific agents. The system is available under the MIT license at \href{https://github.com/PKU-YuanGroup/OpenAI4S}{github.com/PKU-YuanGroup/OpenAI4S}.