Biology subjects

Lin, J.

Publications and source records attributed to Lin, J..

At least 19 recordsLinked to original sources

bioRxiv2026

Accelerated discovery and optimization of a potent andselective MMP12 inhibitor with the machine learningplatform CAMOLE-1

Machine learning techniques are being readily utilized to accelerate drug discovery pipelines. Peptide therapeutics are an emerging modality that offer the advantage of increased potency with reduced toxicity that are amenable to machine learning discovery and optimization. This work describes the Calico All-Atom Molecule Engine (CAMOLE-1), a machine learning platform for de novo peptide inhibitor design. CAMOLE-1 incorporates an all-atom diffusion model that enables rapid non-canonical peptide discovery and optimization, and integrates physics based computational filtering, and biochemical testing. We applied CAMOLE-1 towards the discovery of peptide inhibitors of matrix metalloproteinases (MMPs), a target class that has essentially evaded therapeutic development. Starting from canonical amino acid-based design, an MMP12 peptide inhibitor was identified and further optimized to improve potency. From these initial peptides, we used our all-atom diffusion model to determine a predicted structure, allowing us to design a potent and selective inhibitor over the course of about one year and 270 synthesized peptides. Predicted structures of both canonical and non-canonical peptide inhibitors showed high fidelity to our experimentally determined peptide-bound crystal structures. To our knowledge, this is the first described effort to use machine learning techniques to iteratively drive peptide inhibitor design beyond initial hit discovery into the development of selective, potent lead peptide inhibitors with improved pharmacokinetic properties. Our study illustrates the power of machine learning-driven design of peptide inhibitors, an exciting emerging therapeutic modality.

biochemistry↗

bioRxiv2026

Cortex-wide single-cell activity mapping in freely moving mice

How the cortex supports complex, naturalistic behavior can be revealed only in freely moving animals. Although head-mounted microscopes promise access to the neural dynamics underlying such behavior, fundamental trade-offs among field of view, spatial resolution, device weight, and background contamination have prevented cortex-wide single-neuron recording, leaving distributed cortical coordination largely inaccessible. To meet these challenges, we introduce COMET (Cortex-wide Observational Miniature Epifluorescence Technique), a 3.75-g head-mounted mesoscope that integrates an aberration-controlled wide-field optical design, uniform high-efficiency excitation and single-coaxial power-and-data transmission with an end-to-end denoising and signal-extraction pipeline. This enables cortex-wide calcium imaging at single-cell resolution in freely moving mice. COMET captures activity from >15,000 layer 2/3 neurons per 5-min recording over fields of view up to 10 mm in diameter while preserving natural behavior. Benchmarking against benchtop two-photon microscopy demonstrates high spatial concordance of detected neurons and matched event-level calcium dynamics. We apply COMET to social interaction to quantify region-specific engagement and to an auditory trace fear-conditioning task that requires natural movement and behavioral expression, revealing time-ordered changes in regional firing-rate structure and functional connectivity preceding a global conditioned response. COMET provides an open, lightweight hardware-software framework for linking single-neuron activity to cortex-scale network dynamics during freely expressed behaviors under changing internal or external conditions.

neuroscience↗

bioRxiv2026

A microbiota-driven neuroepithelial feed-forward circuit drives gastroesophageal junction tumorigenesis

The gastroesophageal junction (GEJ) is continuously exposed to refluxed bile acids and microbial perturbations, yet how these environmental cues are converted into persistent oncogenic signaling is unknown. Here we show that bile acid remodeling enriches the oral commensal Prevotella melaninogenica at the GEJ, where bacterial succinate activates both vagal sensory neurons and SUCNR1-expressing tuft cells. Activated tuft cells release 15-deoxy-12,14-prostaglandin J2 (15-d-PGJ2), which sensitizes sensory neurons and sustains nerve remodeling, while neuronal CGRP drives expansion and tuft-cell differentiation of neuro-responsive epithelial progenitors through RAMP1. This self-reinforcing microbiota-neuron-epithelium circuit promotes dysplastic progression, and its genetic or pharmacological disruption, including with the approved CGRP antagonist rimegepant, suppresses tumor development, revealing an interceptable mechanism linking environmental dysbiosis to cancer initiation through neural circuits.

cancer biology↗

bioRxiv2026

Pan-cancer remodeling of reactive oxygen species homeostasis: distinct roles of the ROS-generation and antioxidant arms in immune context and outcome

Reactive oxygen species (ROS) are double-edged in cancer: tumor cells sustain high ROS generation while up-regulating antioxidant defense, so that the net balance between the two arms, rather than absolute ROS abundance, is the biologically meaningful quantity. Whether that balance defines a conserved tumor state across cancer types has been unclear. Here an experimentally measured ROS phenotype, obtained from a three-channel gold-nanocluster array, was generalized to public transcriptomes through a 107-gene ROS program scored in a large cell-line panel and in pan-cancer tumor and adjacent-normal tissue. The optical readout was reproducible and fully discriminative between cell lines, establishing the fingerprint as a quantitative starting point. In tumors, the generation and antioxidant arms were coordinately up-regulated relative to normal tissue, yet the two arms dissociated in their biology. The composite balance score tracked reduced immune infiltration consistently across cohorts, but arm-level decomposition showed that it was carried by low generation-arm activity in immune-cold tumors and followed arithmetically from how the balance score is defined; the association does not implicate antioxidant defense in immune infiltration. The clinical association pointed the other way: adverse outcome was attributable to the antioxidant arm, whereas generation was null. In cohort-held-out validation, tumors were separated from adjacent normal tissue, with redox programs outperforming generic phenotypic programs, indicating that the same program transfers across tumor types. Tumor remodeling of ROS homeostasis is therefore widespread, but its two arms carry distinct meanings: the balance score, driven by its generation term, reports immune context, whereas the antioxidant arm reports clinical risk.

cancer biology↗

bioRxiv2026

GENIE: a multi-agent system for closed-loop autonomous microscopy

Microscopy is foundational to biology, enabling direct observation of otherwise inaccessible cellular structures and processes. Artificial intelligence has advanced microscopy through acquisition control, image analysis, and natural language instrument operation. No system has unified these steps into a complete experimental loop. We introduce an autonomous multi-agent system that turns plain language goals into measurements and statistically tested answers, gating and executing every action in deterministic code. This sets the groundwork for extending microscopy from biological observation to autonomous investigation.

biophysics↗

bioRxiv2026

Structural variants from donor-specific assemblies affect recurrent chromatin contact divergence in humans

Structural variants (SVs) are a major source of human genetic diversity, yet their contribution to three-dimensional (3D) genome organization remains difficult to resolve using a standard reference genome. Here, we integrate haplotype-resolved chromatin interaction maps across 354 human and 12 nonhuman ape haplotypes with matched genome assemblies and long-read transcriptomes to determine how inherited SVs influence chromatin organization. We observe significant improvements in contact mapping, with 13.2% of cis Hi-C contacts showing >25 kbp changes in inferred genomic separation and an additional 2.8% of valid contact pairs recovered when using donor-specific assemblies instead of standard references. We show that haplotype-specific chromatin differences are highly selective rather than pervasive, concentrating within 1,790 recurrent contact difference hotspots associated with a restricted subset of SVs. We identified 2,495 chromatin-structure-associated SVs that are enriched in regulatory genomic contexts and are linked to expression variation. Notably, more than 90% of these SVs are derived in the human lineage, consistent with recent remodeling of regulatory architecture at the population level. These findings highlight the importance of incorporating haplotype diversity from pangenomes into studies of chromatin organization and provide a framework for understanding how inherited genomic differences are linked to functional consequences across populations and evolution.

genomics↗

bioRxiv2026

CHAIR an antigen-resolved antibody-reactome resource for healthy adult immune aging

Antigen-recognition histories accumulate across adult life, but existing human aging resources largely emphasize host-state measurements rather than antigen-resolved humoral memory. We present CHAIR, the Chinese Healthy Aging Immune Resource, which integrates self-antigen, human virome and toxin/virulence-factor PhIP-seq with routine clinical measurements and T/B/NK phenotyping in clinically screened adults aged 20-94 years. The resource framework includes 1,008 unique PhIP-seq samples, 995 clinical records, 977 T/B/NK records and 853 participants with all three layers. Across modeled spaces comprising 23,584 self-antigen targets, 210 virome targets and 1,258 virulence-factor source targets, most measured antibody-reactivity architecture was consistent with broad preservation within domains, whereas age- and sex-associated signals were concentrated in focal targets and modules. Together, CHAIR provides a traceable, antigen-resolved reference framework based on clinically screened adults for investigating focal humoral remodeling across healthy adult aging.

immunology↗

bioRxiv2026

KRAS inhibition drives adaptive surfaceome remodeling and reveals ALPP/ALPPL2 as a targetable vulnerability in pancreatic cancer

KRAS is mutated in over 90% of pancreatic ductal adenocarcinoma (PDAC) and is required for tumor initiation and maintenance. While clinical trials testing mutant-selective and pan-RAS inhibitors (KRASi) have demonstrated significant clinical benefit, genetic and non-genetic mechanisms of resistance limit the long-term efficacy of monotherapy strategies. Here, we used quantitative temporal analysis of the global, phosphorylation, and surface proteome to comprehensively identify non-genetic adaptive pathways and cell surface biomarkers in response to acute and long-term KRAS inhibition in PDAC. Through integrative machine learning analysis of the KRASi-proteome and publicly available datasets, we discovered top priority cell surface targets for combination therapy with KRASi. We identified Placental Alkaline Phosphatase (ALPP) as upregulated in response to KRASi through compensatory JAK2/STAT3 signaling. We developed a novel clinical grade ALPP/ALPPL2 (ALPG)-targeting antibody drug conjugate (ADC) with site-specific MMAE conjugation and demonstrated single agent efficacy in PDAC models, as well as enhanced internalization and anti-tumor efficacy in combination with KRASi. Together, these studies comprehensively define the KRASi-induced proteome and identify ALPP/ALPPL2 as a promising cell surface target for combination with KRASi to enhance response and forestall KRASi resistance.

cancer biology↗

bioRxiv2026

Deciphering Mechanistic Signatures in Drug-Drug Interactions with Dual Topology Graphs

Drug-drug interactions (DDIs) represent a critical challenge in drug development and clinical practice, as they can lead to severe adverse effects, including toxicity and reduced therapeutic efficacy. Deep learning methods have shown promise in large-scale, rapid DDI prediction; however, current approaches suffer from significant limitations in providing mechanistic insights into these interactions. Here, we propose DualTopoDDI, a dual-topology-enhanced interpretable deep learning model for DDI prediction. We applied DualTopoDDI to predict 9.2 billion potential interactions among approved drugs, achieving 97.99% high-confidence predictions. The model demonstrates molecular-level interpretability, identifying key substructures responsible for drug actions in terms of both atom-centric and bond-centric views. DualTopoDDI also excels in elucidating particular DDI toxicity mechanisms, which is validated by its successful explanation of the controversial cardiac toxicity in two COVID-19 drug combination regimens at the time. Evaluations across 11 benchmark datasets demonstrates that DualTopoDDI not only achieves state-of-the-art performance but also showcases robust generalizability and good interpretability. Overall, DualTopoDDI offers a powerful, interpretable tool for understanding and predicting drug-drug interactions, providing critical insights for drug safety and design.

bioinformatics↗

bioRxiv2026

Atlas-scale single-cell analysis beyond in-memory paradigm with scAtlasPy

Single-cell atlases are rapidly outgrowing the memory capacity of standard workstations, challenging the in-memory paradigm underlying mainstream computational ecosystems. Here, scAtlasPy decouples scale of atlas from memory capacity by leveraging the disk-resident computing. It enables full-resolution analysis of a 100-million-cell atlas with only 42.9 GB peak memory, whereas state-of-the-art platforms are limited at 3 million cells with 512 GB memory. scAtlasPy achieves 137,745 cells/s, 10.4x faster than scDataset with 82.6% lower memory usage for random minibatch retrieval. Its extensible architecture offers a flexible platform for diverse atlas-scale analytical tasks, facilitating the discovery of complex cellular heterogeneity and functions in massive cell atlases.

bioinformatics↗

bioRxiv2026

Universal modulation of liquid-liquid phase separation by macromolecules

Liquid-liquid phase separation (LLPS) widely organizes the cellular interior, yet how third-party macromolecules modulate LLPS remains a puzzle. While one would expect that macromolecules can either promote or suppress phase separation, empirical observations show that a small dosage of macromolecules virtually always promotes phase separation. To resolve this paradox, we study the linear response of phase separation to macromolecule addition. We prove that for macromolecules significantly larger than the solvent molecule, their addition essentially always promotes phase separation, regardless of how they interact with the scaffold protein. Crucially, in the biologically relevant scenario of dilute condensates, the linear responses, including the saturation volume fraction, scale proportionally to [1-v2(1+{chi}12)2], where v2 is the macromolecule size and {chi}12 is the scaffold-macromolecule interaction parameter. Importantly, this universal behavior extends to multicomponent solutions in which the effective interaction parameter is composition-dependent. Our work shifts macromolecular modulation of LLPS from empirical observation to predictable design.

biophysics↗

bioRxiv2026

Interface-mediated secondary phase separation

Inside living cells, many types of biomolecular condensates coexist and interact. Recent experiments have shown that new phases often form at the surface of preexisting condensates. However, the mechanisms for this interface-mediated phase transition remain elusive despite its importance to numerous biological processes. Here, we show that interface-mediated secondary phase separation is a universal pathway for forming a new phase in multicomponent solutions. Using Cahn-Hilliard simulations, we successfully generate puncta of client protein on the surface of scaffold condensates. Based on a quasistatic protocol, we theoretically demonstrate that an interfacial instability triggers new-phase formation and predict the amount of client protein needed for the instability to occur. Remarkably, our quasistatic theory successfully predicts the onset of secondary phase separation even if the client is rapidly added or both the client and scaffold are rapidly added. In the latter case, coarsening of the primary condensates drives secondary phase separation. Our work reveals that cells can exploit existing condensates to lower nucleation barriers of new phases, with important implications for protein aggregation.

biophysics↗

bioRxiv2026

TigerAI: An AI-powered genetic evidence platform to support clinical development

Genetic evidence is a major determinant of clinical success in drug development, yet its aggregation has long relied on laborious human curation. Large language models (LLMs) have the potential to rapidly synthesize knowledge across biomedical resources, providing a route to scalable AI-driven genetic evidence generation. Here we develop a novel domain-grounded instruction framework to systematically evaluate GPT-5 for producing genetic evidence relevant to clinical trial success. Using 13,022 target-indication pairs from a comprehensive drug development database, we benchmark LLM-derived evidence against a recent exhaustive human expert-curated study. We find that GPT-5 yields genetic evidence that is at least as informative as expert curation for inferring clinical success, while substantially expanding coverage relative to traditional curation resources. Building on these results, we introduce TigerAI (https://tigerai.bio/), a dual-purpose platform for AI-powered genetic evidence that (i) benchmarks emerging state-of-the-art LLMs and (ii) provides an accessible service for querying reliable AI-generated genetic evidence. These contributions outline a practical, domain-grounded pathway for integrating AI-powered genetic evidence into drug development pipelines and for realizing the potential of LLMs to inform clinical success.

genetics↗

bioRxiv2026

Cross-species discovery of large structural variants reveals distinct evolutionary dynamics in primates

Large-scale structural variants (SVs) are major drivers of genome evolution and disease susceptibility, shaping lineage-specific gene content, genome architecture, and phenotypic diversity. Yet, their systematic detection across divergent species has been hindered by complex rearrangements and reference bias, limiting our understanding of their evolutionary impact. Here we present LGvar, a divergence-aware tool for identifying SVs in population-scale and cross-species genome assemblies, enabling up to a 7-fold increase in the detection of large deletions, insertions, inversions and complex rearrangements ([≥]10 kbp). Application of LGvar uncovers 112 previously undetected large SVs overlooked by conventional methods and quantifies the effect of evolutionary divergence on SV discovery across primate euchromatic genomes, revealing an approximately 30% overall decline in detection sensitivity with increasing phylogenetic distance. Mapping SVs from comparable syntenic regions onto the primate phylogeny shows that large deletions ([≥]10 kbp) accumulate approximately two-fold faster than insertions and eight-fold faster than inversions, in contrast to the dynamics of smaller SVs (<10 kbp). Integrating structurally divergent regions with the synteny region analysis reveals over 606 previously unreported gene gains and losses in great apes and refined the totals to 134 human-lineage-specific and 51 great-ape-lineage-specific protein coding genes. These results suggest that large SVs follow distinct evolutionary trajectories, with potential selective constraints shaping genome structure. Our study establishes a generalizable strategy for cross-species SV discovery and provides a high-resolution view of how large SVs contribute to genome evolution in primates.

genomics↗

bioRxiv2026

An intact primordial interferon antiviral signalling network in amphioxus illuminates the basal chordate origin of vertebrate IFN immunity

The evolutionary origin of the complete interferon (IFN) antiviral signalling cascade across chordates has long remained elusive, as fully functional IFN machinery had not been biochemically reconstituted in basal cephalochordates. Here we combine phylogenetics, AlphaFold3 structural prediction, multi-omic profiling and a full panel of in vitro and in vivo functional assays to characterize an intact primordial IFN network in amphioxus Branchiostoma japonicum, the extant sister group of all vertebrates. Amphioxus encodes a pair of undifferentiated BjTBK1/IKK{varepsilon} paralogs that form cytoplasmic heterodimers, whose conserved kinase-domain phosphorylation residues are mandatory for downstream IRF-driven IFN promoter activation. Despite minimal primary sequence homology with vertebrate IFNs, BjIFN1/2 possess compact -helical core folds that hint at potential structural analogy to vertebrate mucosal type III IFN-{lambda}. We identify two primitive class II cytokine receptors BjCRFB1 and BjCRFB2 that act as functional IFN sensors. Phylogenetic and tertiary structural validations confirm these undiversified ancestral receptor prototypes arose before vertebrate IFN receptor subfunctionalization, and their distinct temporal induction profiles upon viral challenge closely mirror the mucosal immune surveillance orchestrated by vertebrate IFN-{lambda}-IFNLR signalling. AlphaFold3 ligand-receptor docking identifies BjIFN2-BjCRFB2 as the optimal binding pair among all tested combinations, yet all modelled complexes yield low interface scores indicative of weak ancestral intermolecular interactions. Downstream signal transduction relies on two STAT paralogs, BjSTATa and BjSTATb, which jointly mediate transcriptional activation of core antiviral effector genes. The conserved GTPase BjMx serves as a central IFN-stimulated effector to suppress viral replication and maintain tissue immune homeostasis. Collectively, our structural and functional evidence demonstrates that the full hierarchical IFN signalling apparatus was fully assembled in basal cephalochordates prior to vertebrate radiation. The amphioxus IFN cascade preserves core ancestral molecular traits including structure-dependent ligand conservation and primitive low-affinity ligand-receptor interactions. This study resolves a long-standing evolutionary gap and provides definitive functional evidence demonstrating that IFN-mediated innate immunity constitutes an ancestral chordate trait, rather than a vertebrate-specific evolutionary innovation. It also puts forward the hypothesis that mucosal surveillance may represent the ancestral mode of chordate IFN defence.

evolutionary biology↗

bioRxiv2026

Biased stochastic motor dynamics drive bidirectional, processive translocation by the AAA+ disaggregase Hsp104.

Protein aggregation disrupts proteostasis and drives neurodegeneration. Hsp104 is a hexameric, ring-shaped AAA+ ATPase that dissolves protein aggregates, yet how hexamers translocate and extract polypeptides trapped in mechanically resistant aggregates remains unclear. Using substrates that recapitulate the physical constraints of aggregates, we establish that Hsp104 is a processive, bidirectional translocase that can dynamically switch direction while threading a single polypeptide. On mechanically restrained substrates and prions, Hsp104 hexamers execute biased stochastic transitions among three conformational states at individual interprotomer interfaces: closed, extended, and a previously unobserved hyperextended form. These transitions follow kinetically favored paths rather than a rigid rotary sequence. The resulting biased stochastic stepping, enabled by the conformational plasticity of Hsp104 hexamers, underpins operational adaptability and redefines the functional logic of AAA+ motors.

biochemistry↗

bioRxiv2026

Pangenome discovery and characterization of human protein-coding duplicated genes

Protein-coding genes mapping to high-identity segmental duplications (SDs) have been difficult to annotate and characterize and are the source of most previously unknown protein-coding genes being discovered as part of the human pangenome. Here, we combine long-read assembled human genomes (298) and long-read transcriptome data (5.6 billion full-length cDNA from 83 tissues) to phylogenetically interrogate 493 gene families discovering 2713 potentially copy number polymorphic genes not present in the human reference genome. For reference SD gene families where paralog specificity can be assigned, we find that 60.0% are expressed and maintain open reading frames, with 45.7% showing high expression in brain, embryo, or testis. We revise 386 gene models, including 150 that absent or different from current T2T-CHM13 gene annotation and 236 (35.1%) pseudogenes as protein-coding where we find evidence of transcription, an open reading frame, and chromatin-accessible promoters. We find that 24.2% of SD genes show evidence of constraint for both copy number and amino acid mutation. The majority of these constraint genes are ancestral, whereas only 16.2% of derived duplicated genes that emerged recently in the human lineage show evidence of constraint. The pangenome provides unparalleled specificity to understand genetic variation in SD genes allowing us to distinguish functional genes from pseudogenes and highlighting potential gene innovations that arose most recently in human evolution.

genomics↗

bioRxiv2026

Machine learning-assisted directed evolution of plant Rubisco

Ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) is foundational to life on Earth, catalyzing carbon dioxide (CO2) fixation to generate biomass. However, Rubisco is a slow and inefficient enzyme that has proven challenging to engineer. We applied the structure-informed machine learning (ML) model ESM-IF1 to identify plausible amino acid sites in the large subunit of Nicotiana tabacum Rubisco to target for directed evolution. ML-assisted library design followed by selection in Rubisco-dependent Escherichia coli identified multiple enriched variants displaying improved catalytic efficiency. Several improved variants carried amino acid changes not found in the evolutionary lineage of plants, despite being assembly competent in plant chloroplasts, demonstrating that ML-assisted protein design can explore functional sequence space beyond what is observed from natural sequence diversity. Most prominently, the T391I substitution improved carboxylation rate by 29% and aerobic carboxylation efficiency by 43%. Our findings demonstrate the utility of ML-assisted evolution for engineering Rubisco with improved carboxylation efficiency and potential for enhancing crop productivity.

plant biology↗