Biology subjects

Su, H.

Publications and source records attributed to Su, H..

At least 19 recordsLinked to original sources

bioRxiv2026

High-plex protein imaging across thick tissue by synchronized amplification

High-plex protein imaging is essential for decoding complex tissue organization, yet it remains largely confined to thin sections. Current approaches in millimeter-thick samples suffer from poor signal penetration and depth-dependent, uneven signal amplification. Even with specialized instrumentation to facilitate probe diffusion, iterative millimeter-thick tissue workflows can take more than four days per staining round, and reported eleven-marker experiments extend beyond a month. We introduce Amplification by Cyclic Extension (ACE), a DNA-barcoded, synchronized in situ amplification strategy that uses thermally controlled cyclic extension to coordinate DNA concatemer growth across tissue depths. By synchronously activating and coordinating reaction progression throughout the sample, ACE enables controlled, uniform signal amplification across tissue depths. Using ACE, we demonstrate 63-plex protein and RNA imaging in thin sections, 22-plex protein imaging in 200 m mouse brain sections, and 16-plex protein imaging in 1 mm-thick brain tissue encompassing over three million cells. For millimeter-thick tissues, ACE reduces per-round target-switching time from multiple days to less than eight hours, shifting volumetric high-plex protein imaging from months to days. Together, these results establish ACE as a synchronized amplification strategy that enables scalable high-plex protein imaging in thick tissues.

bioengineering↗

bioRxiv2026

Neural signatures of spontaneous transitions between internal and external thought

The human mind constantly shifts between internal representations and the external environment, yet the neural mechanisms underlying such spontaneous transitions remain underexplored. Here, we analyzed a think-aloud functional magnetic resonance imaging dataset, in which participants continuously verbalized their thoughts, to identify neural activity predicting transitions between internally and externally oriented thought. Internal-to-external transitions were preceded by increased activation in the salience/ventral attention network, with the strongest effect observed in the right temporoparietal junction. The spatial pattern of this pre-transition activation was positively associated with acetylcholine receptor density, suggesting a role for cholinergic signaling in cognitive reorientation. Pre-transition activation was itself preceded by a large-scale brain state proposed to serve as a flexible hub between functionally specialized states, indicating that spontaneous transitions are more likely when the brain occupies this intermediate configuration. Together, these findings suggest that multilevel neural mechanisms support flexible reorientation along the internal-external dimension of spontaneous cognition.

neuroscience↗

bioRxiv2026

NAE1-Dependent Protein Neddylation Preserves Endothelial Identity and Vascular Integrity

Background: Endothelial dysfunction is a central driver of cardiovascular and inflammatory diseases, yet the post-translational mechanisms that preserve endothelial homeostasis remain incompletely understood. Protein neddylation, the covalent conjugation of a ubiquitin-like modifier, regulates diverse cellular processes, yet its physiological role in the vascular endothelium remains unknown. This study investigated whether protein neddylation is required to preserve endothelial identity and vascular homeostasis. Methods: We generated tamoxifen-inducible endothelial-specific Nae1 knockout mice to inhibit neddylation and combined bulk RNA sequencing, single-cell and single-nucleus transcriptomics, quantitative proteomics, biochemical analyses, and gain- and loss-of-function approaches to define the role of endothelial neddylation in vascular homeostasis and inflammatory injury. Results: Endothelial-specific Nae1 deletion caused rapid mortality associated with vascular leakage, platelet accumulation, inflammation, and multi-organ injury. Multi-omics analyses demonstrated profound loss of endothelial identity, characterized by suppression of core endothelial programs and activation of inflammatory, procoagulant, and pyroptotic pathways. Single-cell analyses revealed progressive endothelial dysfunction culminating in depletion of the endothelial population and remodeling of the vascular niche. Mechanistically, endothelial neddylation deficiency activated gasdermin D (GSDMD)- and gasdermin E (GSDME)-dependent pyroptosis, whereas dual inhibition of GSDMD and GSDME markedly attenuated inflammatory transcriptomic remodeling, vascular injury, hepatocyte death, immune cell infiltration, and platelet accumulation. Translational analyses demonstrated reduced endothelial neddylation in experimental endotoxemia and decreased expression of neddylation pathway components in human atherosclerosis and COVID-19 datasets. Conversely, restoration of endothelial neddylation partially reversed inflammatory endothelial transcriptomic reprogramming in vivo. Conclusions: NAE1-dependent protein neddylation is an essential regulator of endothelial identity and vascular integrity. Loss of endothelial neddylation promotes gasdermin-dependent pyroptosis and thrombo-inflammatory vascular injury, whereas restoration of the neddylation pathway mitigates inflammatory endothelial dysfunction. These findings identify endothelial neddylation as a fundamental mechanism maintaining vascular homeostasis and a potential therapeutic target for cardiovascular and inflammatory diseases.

pathology↗

bioRxiv2026

NACraft: Programmatic nucleic-acid aptamer design via all-atom structure-model feedback

Protein-nucleic-acid interactions underpin diverse biological processes and provide a basis for molecular sensing, regulation and therapeutic intervention. However, the coupled dependence of aptamer function on nucleotide sequence, three-dimensional folding and target binding makes rational RNA and DNA binder design challenging. Here we present NACraft, a training-free and programmatic framework for all-atom nucleic-acid aptamer design based on backpropagation through structure-model feedback. By composing binding, sequence-similarity and anti-binding constraints, NACraft supports de novo generation, similarity-guided sampling and target-selective design within a unified optimization framework, without task-specific training or fine-tuning. Computational experiments showed that NACraft generated high-confidence candidates de novo across diverse protein targets, with further improvements achieved through similarity-guided design for both RNA and DNA complexes. Its target-selective design capability was further validated in silico, with 69.44% of paired candidates generated to favour the positive target EGFR over the off-target HER2. Under matched independent AlphaFold3 evaluation, NACraft achieved better performance than ODesign in 10 of 11 NA-12 targets and 17 of 20 protein target-length settings. Together, these results demonstrate the effectiveness and versatility of NACraft and extend structure-model hallucination toward programmatic nucleic-acid aptamer design. Codehttps://github.com/OTEAM-AI4S/NACraft

bioinformatics↗

bioRxiv2026

DrtA, a novel major facilitator superfamily transporter, contributes to intrinsic tolerance to the chemotherapeutic agent mitomycin C in Acinetobacter baumannii

Human-targeted non-antibiotic drugs are increasingly recognized for their intrinsic antibacterial activity, yet Gram-negative pathogens such as Acinetobacter baumannii exhibit substantial tolerance to these compounds. This tolerance is largely attributed to restricted outer membrane permeability and the activity of multidrug efflux systems. While Resistance Nodulation Division (RND) transporters have been extensively studied, the contribution of individual Major Facilitator Superfamily (MFS) transporters to non-antibiotic drug tolerance remains poorly understood. Here, we investigated H0N29_04330, designated Drug Resistance Transporter A (DrtA), a Bcr/CflA subfamily MFS transporter, to define its substrate specificity and contribution to antibiotic and non-antibiotic drug tolerance. DrtA was highly conserved across the A. calcoaceticus-baumannii complex and exhibited broad substrate specificity when heterologously expressed in an efflux-deficient Escherichia coli background, conferring resistance to benzalkonium, ethidium bromide, phenicols, and the antineoplastic agent mitomycin C. Intriguingly, drtA expression increased E. coli susceptibility to the antifolate compounds methotrexate and aminopterin, suggesting that DrtA may recognize folate-related metabolites rather than function as a dedicated antifolate transporter. In contrast, loss of drtA in its native A. baumannii host primarily impaired tolerance to mitomycin C, highlighting a context-dependent physiological role influenced by the extensive functional redundancy among A. baumannii efflux systems. Site-directed mutagenesis further identified M18 and the membrane-embedded protonatable residue D26 as critical determinants of DrtA transport activity and substrate recognition. Together with previous characterization of CraA, our findings demonstrate that Bcr/CflA subfamily MFS transporters contribute to protection against structurally diverse human-targeted compounds and expand the functional landscape of efflux-mediated intrinsic tolerance beyond conventional antibiotic resistance.

microbiology↗

bioRxiv2026

Ependymomas are cancers of the pre-neural crest/roof plate lineage

Distinct molecular variants of the brain cancer ependymoma are distributed along the rostral-caudal extent of the central nervous system (CNS). Historically proposed to arise from ventricular ependyma, recent studies have suggested conflicting cellular origins, including the neural radial glia and the roof plate lineages. Using single-cell transcriptomics, immunohistochemistry, and lineage tracing, we demonstrate that ependymomas across all CNS compartments transcriptionally mirror MSX1+ve pre-neural crest/roof plate (Pre-NC/RP) lineage derivatives. Ependymoma subgroups recapitulate the spatial and molecular diversity of regional Pre-NC/RP populations, while retaining conserved MSX1 expression. Expression of the oncogenic fusion ZFTA-RELA within the murine Pre-NC/RP lineage generated tumors that faithfully resembled human ependymoma. These findings identify a common embryonic cellular origin for ependymomas and reconcile previously conflicting models of tumorigenesis.

developmental biology↗

bioRxiv2026

VigExp: A functionally verified platform for aiding cowpea (Vigna unguiculata) and related legume crop improvement

Legumes include some of the worlds most significant crop species, such as cowpea (Vigna unguiculata), a subsistence crop widely grown in sub-Saharan Africa. Despite their importance, legume crop improvement is hindered by a lack of high-resolution expression data, particularly for reproductive tissues and cell types. Here, we report on VigExp, a tool for visualising cowpea gene expression datasets. We demonstrate its utility across a range of vegetative and reproductive cell types of varieties IT97K-499-35 and IT86D-1010, which exhibit 93.75% protein sequence conservation and are amenable to stable transformation. This includes previously published transcriptomes of vegetative, floral and seed tissues, combined with developmentally staged male and female reproductive tissues. Also integrated are novel transcriptomes of laser-captured cell types covering reproductive development from meiosis to early embryo formation post-fertilisation. Spatial expression patterns and transcript levels can be visualised through an electronic fluorescent pictograph (eFP) browser. Validated by RT-qPCR, in situ hybridisation, transgenic, and CRISPR gene editing analyses, the predictive accuracy of VigExp matches prior cowpea functional study observations. Critical genes for nodule development and regulation were also identified and their expression patterns established in cowpea. Novel reference genes, constitutively expressed gene promoters for visualisation markers/gene-editing, and tissue- and cell-specific gene promoters for targeting these regions, were identified. The A-type cyclin, VuTAM2, was also identified, with a critical role in male meiosis established. Collectively, VigExp represents an adaptable and updatable resource to support crop improvement in cowpea and other legumes, which are often highly syntenic with respect to genome composition.

plant biology↗

bioRxiv2026

Experiment-free learning of exoskeleton assistance is not an unsolved problem

We present three quantitative methods: 1) estimation of exoskeleton mechanical power and energy ratio from published data, 2) a systematic review of the exoskeleton literature on reported energy ratios, and 3) timing correction analysis of the replication experiment, to address concerns raised by Collins et al. (2026) about Luo et al. (2024). Together, these analyses support the reported metabolic reductions and the validity of exoskeleton control via learning in simulation. The critique rests on an unsupported premise: that exoskeleton energy ratios above 4 are physiologically implausible. This premise of Collins et al. (2026) is not supported by the cited evidence, and the error originates in their own cited source. Sawicki and Ferris (2009), the paper they invoke as authority for the limit of 4, state explicitly that "reported values of the muscular efficiency range from 0.10 to 0.34, with many sources assuming an average of [~]0.25." The value of 4 corresponds to this average, it is not a physiological ceiling. Treating an average as a physiological upper limit is a fundamental error. The published exoskeleton literature further contradicts the claim, including work by the authors of the critique themselves (Collins et al., 2015: 4.3; Young et al., 2017: 5.0) and independent work (Malcolm et al., 2013: 4.8; Seo et al., 2017: 6.7). In contrast, our walking energy ratio is 2.4, calculated directly from Fig. 4 of our paper. Our device delivers higher peak torque (14.1 Nm vs. 10.9 Nm, Lim et al., 2019) and achieves a slightly larger metabolic reduction (24.3% vs. 21%). Independent groups have since demonstrated meaningful metabolic reductions using learning-in-simulation frameworks, including Barati et al. (2026, 15.2% mean and 22.5% maximum) and Zhou et al. (2025, [~]20% during running). The claim of Collins et al. (2026) that this problem "remains unsolved" is directly contradicted by these independent results. The experiment in the critique is not a valid replication of our method. Our controller is a neural network with [~]10,000 parameters learned through deep reinforcement learning in musculoskeletal simulation; the critique instead applies a pre-programmed fixed torque curve with no learnable parameters. Beyond this, the replication contains three methodological errors: 1) a heel-strike timing assumption producing offsets up to 30% of the gait cycle; 2) an averaged torque profile that discards subject-specific control; and 3) a device [~]50% heavier than ours (4.8 kg vs. 3.2 kg) without measuring the metabolic penalty of the added weight. The critique also misreports Samsung data, with reported values approximately double those in the original publication, errors that directly underpin their physiological limit argument.

bioengineering↗

bioRxiv2026

Kaposi Sarcoma herpesvirus reactivation and replication are dispensable for cell-lineage specific tumorigenesis in whole genome transgenic mice

Herpesvirus reactivation and replication depend on an ordered cascade of gene expression. The viral immediate-early replication and transcription activator protein (Rta) is necessary and sufficient to initiate this process. For Kaposi sarcoma herpesvirus (KSHV), Rta-interacting and Rta-regulated proteins are thought to contribute to cellular transformation. Using a novel, whole-genome transgenic mouse model, we show that KSHV tumorigenesis is independent of Rta in vivo. Rta-deleted, KSHV bacmid transgenic (tgKSHV{Delta}Rta) mice developed latency-associated nuclear antigen (LANA)-positive vascular sarcomas with the same endothelial cell-lineage composition and transcriptional profile as human Kaposi sarcoma (KS). The viral miRNAs were expressed in the tumors, as were all canonical latency and the K1 and K15 oncogenes. Importantly, this system is the first to model both KS and lymphoma, reflecting the development of both malignancies observed in KSHV-infected individuals. These results separate virus replication from KS tumorigenesis, which can proceed independently so long as the virus genome is maintained in a susceptible cell. SignificanceKaposi sarcoma (KS) is the most common cancer in people living with HIV worldwide. Like all herpesviruses, KS-associated herpesvirus (KSHV) has a latent and lytic phase. Hitherto, viral lytic genes are thought to play a role in KS. The KSHV replication and transcription activator (Rta) is both necessary and sufficient for reactivation and lytic viral replication, but Rta was not required for vascular tumor development in a whole virus genome mouse model. This new tgKSHV{Delta}RTA mouse model separates virus replication from viral tumorigenesis and defines a minimal set of viral oncogenes. This model enables mechanistic studies in vivo, and treatment modalities to be evaluated within a complete immune and stromal microenvironment.

Cancer Biology↗

bioRxiv2026

Dual-Stream Compression of High Bit-Depth Medical Images with Application to DNA Storage

High bit-depth medical images preserve subtle intensity variations that are important for quantitative analysis and clinical interpretation, but their large dynamic range poses challenges for efficient compression. We propose a bit-plane-aware dual-stream compression framework for 16-bit medical images by separately modeling the most significant bit (MSB) and least significant bit (LSB) components. The MSB structural stream is encoded using JPEG coding with a Duplicate Segment Skipping (DSS) strategy to exploit spatial and segment-level redundancy, while the LSB detail stream is compressed using learned image compression to represent residual variations and fine-grained details. Experiments on four MRI and CT datasets show that the proposed method consistently outperforms representative traditional and learning-based codecs, achieving the lowest bit rate across all datasets. Meanwhile, it preserves high reconstruction fidelity. As a downstream application, we further demonstrate that the compressed bitstreams can be effectively integrated with DNA encoding and converted into sequences with favorable biochemical properties.

bioinformatics↗

bioRxiv2026

Expansion Revealing of Pathology Resolves Nanostructures Associated with Inflammatory Phenotypes in COVID-19 Decedent Human Brain Tissue

Expansion revealing (ExR) elucidates cellular organization by separating proteins within dense nanostructures by 20x linear expansion, but requires fixation procedures incompatible with human pathology specimens. Here, we report ExR of pathology (ExRPath), which attains [~]20 nm resolution and decrowding of such tissues, through iterative 20x expansion, adapted to human brain pathology specimens. We also report a single-shot 15x expansion protocol for such tissues (15ExMPath), achieved through one-shot 15x expansion. Applying ExRPath and 15ExMPath to COVID-19-decedent brain tissue reveals periodic amyloid nanoclusters that co-localize with SARS-CoV-2 in a rare minority of patient specimens, pointing to a potential neuroinflammatory phenotype associated with COVID-19, and highlighting the power of high-throughput nanoimaging, empowered by expansion microscopy, for discovering potential novel disease mechanisms.

neuroscience↗

bioRxiv2026

Causally validated phase-amplitude coupling enables high-fidelity motor decoding for next-generation brain-computer interfaces

Modern Brain-Computer Interfaces (BCIs) face a fundamental performance plateau due to their reliance on broad spectral power reduction (Event-Related Desynchronization, ERD) as the primary decoding feature. ERD acts as a coarse metabolic proxy for cortical activation, discarding the high-frequency temporal syntax necessary for high-dimensional motor control. Here, we demonstrate that Phase-Amplitude Coupling (PAC)--specifically, the phase-locking of high-gamma (70-150 Hz) firing to residual Beta oscillations--provides a high-fidelity temporal feature for decoding motor intent. Using high-density electrocorticography (HD-ECoG) during motor imagery, we show that incorporating Beta-PAC into machine-learning classifiers (LDA/SVM) significantly outperforms traditional power-based features. To definitively validate the causal robustness of this feature against mathematically-derived artifacts, we leveraged a rare in vivo human structural lesion model. In a patient with focal tumor infiltration of the motor tract, the metabolic gate (ERD) was preserved, yet structural uncoupling completely abolished Beta-PAC, collapsing network topology and reducing single-trial decoding accuracy to chance levels. Our findings causally validate Beta-PAC as a robust, independent control signal, establishing a physiological foundation for next-generation, phase-dependent neuroprostheses.

neuroscience↗

bioRxiv2026

An fMRI dataset of verbalized spontaneous thought with annotated transcripts and self-report trait measures

Spontaneous thought is pervasive in everyday human cognition, yet datasets capturing its neural dynamics under minimally interrupted conditions remain limited. The current dataset was acquired from a think-aloud functional MRI experiment in which 118 participants continuously verbalized their spontaneous thoughts during 10-minute scanning sessions. The raw MRI data and verbal transcripts with sentence-level timestamps were previously released and analyzed in our prior study examining neural activity associated with thought transitions. Building on that release, we additionally provide preprocessed MRI data, speech transcriptions with word-level timestamps aligned to image acquisition, large language model-generated ratings of transcribed thoughts across emotional and sensory dimensions, and self-report survey measures assessing personality, mental health, and cognitive abilities. Validation analyses demonstrated activation in expected cortical regions associated with speech production and sensory content identified from transcript annotations, agreement between language model and human ratings, and adequate internal consistency of survey measures, supporting the datasets overall quality. This dataset enables reuse for investigations of spontaneous thought, speech generation, and individual differences using naturalistic functional MRI data.

neuroscience↗

bioRxiv2026

Heterotrimeric G protein αi2 Sequesters RasGAP to Control Neutrophil Sensitivity and Chemotaxis

G protein-coupled receptors (GPCRs) direct neutrophil chemotaxis through heterotrimeric G proteins, yet the downstream effectors of the predominant Gai isoform, Gai2, remain incompletely defined. Here we identify the Ras GTPase-activating protein CAPRI (RASA4) as a functional effector of Gai2 that links GPCR signaling to Ras adaptation. Using AlphaFold3-based structural modeling and binding free-energy calculations and experimental verification, we reveal that constitutively active and structurally altered Gai2 mutants (Q205L and T182A) exhibit enhanced binding to CAPRI. Neutrophils expressing these mutants display elevated basal Ras activity, heightened sensitivity to chemoattractant, and improved chemotaxis in low- or subsensitive-concentration gradients. However, these cells exhibit excessive Ras activation and impaired chemotaxis at high, saturating chemoattractant concentrations, while maintaining near-normal responses at intermediate concentrations. These results reveal an upward shift in the concentration range for efficient chemotaxis. Our findings not only establish a previously unrecognized Gai2-CAPRI signaling axis that tunes Ras adaptation but also define a mechanism by which heterotrimeric G proteins calibrate leukocyte navigation across diverse chemoattractant gradients.

immunology↗

bioRxiv2026

Multidimensional Mechanistic Study of Panax Notoginseng Saponins in the Treatment of Alcohol-Induced Osteonecrosis of the Femoral Head: Integrating Network Pharmacology, Molecular Dynamics Simulation and In Vivo Validation

BackgroundAlcohol-induced osteonecrosis of the femoral head (AIONFH) is an orthopedic disorder from chronic alcohol abuse, characterized by disrupted femoral head blood supply, osteocyte death and structural collapse. Current hip-preserving therapy is unsatisfactory, and most patients eventually require total hip arthroplasty. Panax Notoginseng Saponins (PNS), the core active component of Panax notoginseng, exerts pro-angiogenic and anti-osteocyte apoptosis effects, but its specific therapeutic mechanism remains unclear. ObjectiveThis study used network pharmacology, molecular dynamics simulation and animal experiments to identify PNSs active components, core targets and key pathways for AIONFH, verify its in vivo efficacy, and provide a scientific basis for clinical application. MethodsPNS active components, their targets and AIONFH-related targets were screened from databases; intersection targets constructed an interaction network, core targets were screened by three machine learning algorithms, with concurrent GO and KEGG analysis. Molecular docking was performed between core targets and PNS components; Gromacs 2022 conducted 100 ns simulation to evaluate complex stability. AIONFH rat models were grouped with 4-week intragastric intervention; pathology, immunofluorescence and PCR were used for detection. Results and DiscussionNetwork pharmacology identified 127 PNS targets and 18 intersections with 672 AIONFH targets. Six core targets (including FGF2, HSD11B1) were screened; KEGG indicated VEGF pathway as key. Ginsenoside Re bound HSD11B1 with the lowest binding energy (-12.4 kcal/mol), and 100 ns simulation confirmed complex stability. Animal experiments showed PNS improved trabecular structure and regulated osteocyte activity. PNS treats AIONFH via multi-component, multi-target mode, core mechanism being osteocyte apoptosis inhibition. Results and DiscussionNetwork pharmacology screening identified 127 potential targets of PNS, and 18 potential intersection targets were obtained by overlapping with 672 AIONFH-related targets. Six core targets including FGF2 and HSD11B1 were screened out by machine learning, and KEGG analysis indicated that the VEGF pathway and other pathways were the key signaling pathways for PNS action. Molecular docking showed that Ginsenoside Re had the lowest binding energy with HSD11B1 (-12.4 kcal/mol), and 100 ns molecular dynamics simulation confirmed the stable conformation of this complex. Animal experiments demonstrated that PNS could improve trabecular bone structure and regulate osteocyte activity. In summary, PNS exerts a therapeutic effect on AIONFH through a multi-component, multi-target and multi-pathway mode, with the core mechanism of inhibiting osteocyte apoptosis.

pharmacology and toxicology↗

bioRxiv2026

Spikes meet Spins: Quantum-Native Neural Decoding for Ultra_Low-Latency Brain-Computer Interfaces

Brain - computer interfaces (BCIs) require rapid and accurate decoding of neural activity, yet conventional computing architectures face growing latency as neural recording scales. We demonstrate a quantum computing - enabled neural decoding approach using a physical 1000-qubit coherent photonic Ising machine, in which inference is performed through hardware energy relaxation rather than numerical computation. By mapping sparse neural spike patterns onto Ising Hamiltonians, our hardware-native Quantum Semi-Restricted Boltzmann Machine achieves up to 96.2% accuracy across public in vivo datasets spanning multiple species and modalities. We report hardware-verified median latencies of 0.075 ms-- a tenfold speedup over GPUs--with complexity-invariant scaling. These results establish quantum computing as a viable pathway toward ultra-low-latency neural decoding for future BCI systems.

neuroscience↗

bioRxiv2026

MolClaw: An Autonomous Agent with Hierarchical Skills for Drug Molecule Evaluation, Screening, and Optimization

Computational drug discovery, particularly the complex workflows of drug molecule screening and optimization, requires orchestrating dozens of specialized tools in multi-step workflows, yet current AI agents struggle to maintain robust performance and consistently underperform in these high-complexity scenarios. Here we present MolClaw, an autonomous agent that leads drug molecule evaluation, screening, and optimization. It unifies over 30 specialized domain resources through a three-tier hierarchical skill architecture (70 skills in total) that facilitates agent long-term interaction at runtime: tool-level skills standardize atomic operations, workflow-level skills compose them into validated pipelines with quality check and reflection, and a discipline-level skill supplies scientific principles governing planning and verification across all scenarios in the field. Additionally, we introduce MolBench, a benchmark comprising molecular screening, optimization, and end-to-end discovery challenges spanning 8 to 50+ sequential tool calls. MolClaw achieves state-of-the-art performance across all metrics, and ablation studies confirm that gains concentrate on tasks that demand structured workflows while vanishing on those solvable with ad hoc scripting, establishing workflow orchestration competence as the primary capability bottleneck for AI-driven drug discovery.

bioinformatics↗

bioRxiv2026

Fatty acid scavenging enables cancer escape from KRAS inhibition

Although inhibitors of oncogenic KRAS have shown clinical efficacy1, resistance to KRAS inhibition is common2, and its molecular basis remains unclear. Here we show that KRASi-resistant cancer cells sustain mitochondrial bioenergetics through enhanced fatty acid (FA) metabolism, despite suppression of canonical KRAS signaling. Specifically, KRASi-resistant pancreatic cancer cells exploit macropinocytosis to scavenge FA released from adipose tissue, fueling beta-oxidation independently of KRAS-PI3K signaling. This adaptive metabolic program is driven by the adhesion G protein-coupled receptor ADGRB1, which activates non-canonical PI3K{gamma}-PAK1 signaling to stimulate macropinocytosis and maintain metabolic homeostasis under KRASi. Disruption of ADGRB1-PI3K{gamma} signaling dismantles this metabolic program and restores KRASi sensitivity. This pathway operates across multiple KRAS-mutated cancers and is associated with poor therapeutic response and outcome. These findings offer a promising strategy for overcoming KRASi resistance.

cancer biology↗