Biology subjects

Chen, S.

Publications and source records attributed to Chen, S..

At least 19 recordsLinked to original sources

bioRxiv2026

Population-scale dietary signatures of the oral microbiome

Diet is an environmental factor that influences oral microbial communities. How specific foods, nutrients and dietary patterns relate to microbial diversity, species composition and functional potential remains incompletely characterised. Here, we combined detailed dietary records with buccal metagenomic profiles from 5,824 Human Phenotype Project participants. Using these data, we examined diet-oral microbiome associations with prediction models and covariate-adjusted analyses. Plant-based and guideline-based dietary patterns showed similar associations across oral microbial species. However, species within the same genus could show opposite associations with a dietary pattern. For example, a whole-food plant-based diet score was positively associated with Streptococcus sanguinis but negatively with S. gordonii and S. oralis. Some diet-oral microbiome associations were replicated at the genus level in an independent US National Health and Nutrition Examination Survey cohort of 4,667 participants. We map dietary associations with oral microbial diversity, species composition and functional potential, revealing shared microbial signatures across dietary patterns.

microbiology↗

bioRxiv2026

Widespread Presence of Catalytic RNAs in All Domains of Life

Very few ribozymes have been found in complex eukaryotes. Here, we developed a novel method, AbRiDGed, for highly accurate and sensitive ab initio self-cleaving ribozyme discovery. Our results indicate the existence of millions of self-cleaving ribozymes in humans and other species. We characterized ~150 novel classes of self-cleaving ribozymes, including one class which exhibits noncanonical biochemical properties and widespread presence in all domains of life and appears to play specific functions in complex organisms. These results revealed that the true complexity of the "ribozyme-ome" is several orders of magnitude higher than previously estimated. We demonstrate that newly identified self-cleaving ribozymes influence organismal fitness by regulating gene expression and enhancing transcriptome complexity in vivo. Moreover, these ribozymes hold promise as foundational tools for biotechnological applications, such as direct RNA detection technologies. Overall, catalytic RNAs are prevalent and diverse, and may represent a hidden regulatory layer in various organisms.

molecular biology↗

bioRxiv2026

Erythrocyte extravasation drives ferroptosis and inflammatory amplification in eccentric contraction-induced skeletal muscle injury via the heme-HO-1-iron axis

Eccentric contraction-induced skeletal muscle injury is a common sports-related condition characterized by pain, impaired muscle function, slow recovery, and frequent recurrence, yet treatment remains symptomatic and mechanism-based targets are lacking. Although microvascular disruption and erythrocyte extravasation characterize severe muscle strain, the metabolic fate of extravasated erythrocytes has received little attention. In hemorrhagic and hemolytic conditions, erythrocyte-derived heme can activate heme oxygenase-1 (HO-1)-mediated iron release and ferroptosis, potentially amplifying inflammation; however, whether this pathway contributes to eccentric muscle injury remains unknown. To address this gap, we established a parameter-controlled rat model in which the gastrocnemius underwent 0, 50, 100, or 150 eccentric stretch repetitions. Muscle torque, gait, weight-bearing, endurance, pain threshold, histopathology, microcirculatory perfusion, macrophage polarization, and heme, ferrous iron, glutathione (GSH), malondialdehyde (MDA), glutathione peroxidase 4 (GPX4), and pro-inflammatory cytokine levels were assessed. Increasing repetitions produced graded functional impairment, myofiber damage, erythrocyte extravasation, inflammatory infiltration, and microcirculatory disturbance. Heme accumulation was accompanied by HO-1 upregulation in perivascular CD68-positive (CD68) macrophages, muscle iron accumulation, GSH and GPX4 depletion, elevated MDA, and increased M1 macrophage and cytokine levels. These findings suggest a potential heme-HO-1-iron-ferroptosis-associated inflammatory mechanism in eccentric muscle injury, providing a rationale for mechanistic validation and targeted therapeutic development. Keywords: Eccentric contraction-induced muscle injury; Erythrocyte extravasation; Heme oxygenase-1; Ferroptosis; Macrophage polarization

immunology↗

bioRxiv2026

Spontaneous brain-wide rotational dynamics constrain cue-evoked behavior

Uncovering low-dimensional representations of brain-wide neural dynamics and their relationship to behavior remains a central challenge. Here, we show that spontaneous brain-wide rotational dynamics constrain cue-evoked behavior and learning. To systematically establish these findings, we developed a dimensionality-reduction technique called Rotational Component Analysis (RCA) that optimizes the detection of low-variance rotational modes even within individual trials. We applied RCA to motor-cortical spike-train data from monkeys and humans, mouse cortex-wide calcium imaging, and human functional magnetic resonance imaging (fMRI) and magnetoencephalography (MEG). In mice, a rotational mode identified from activity recorded at rest predicts cue-evoked behavior: the phase of rotation when a randomly timed cue arrives predicts whether the animal subsequently pulls the lever. We explain this surprisingly strong correlation using both a recurrent neural network (RNN) and a minimal two-dimensional model of response learning, in which the intrinsic rotational mode serves as a dynamical constraint on response and learning. We further identify double-rotation modes containing two centers of rotation that separate preparatory from task dynamics in both mouse and human data. Remarkably, corresponding modes also organize neural dynamics during a non-motor human task, and rotational irreversibility captures early psychosis even when principal component analysis shows no group difference. Our work establishes rotational dynamics as low-dimensional coordinates that constrain how ongoing brain states are converted into behavior. More broadly, RCA provides a versatile tool for detecting hidden nonreciprocal dynamics in biological and physical datasets.

neuroscience↗

bioRxiv2026

Organelle-Specific pH Quantification Using a Protein-Based Quantum Sensor

Quantum sensing enables measurements of physicochemical changes within living cells at the organelle level. Conventional nanoparticle-based sensors, however, face limitations in subcellular targeting, intracellular tracking, and following dynamic biological processes across distinct intracellular environments. Here, we investigate MagLOV2, a genetically encoded quantum sensor, by examining its organelle-specific localization, magnetic field effects (MFEs) during mitophagy, and pH dependent response kinetics. MagLOV2 retained magnetic responsiveness in the cytoplasm, nucleus, and mitochondria, with fluorescence contrasts of 17.1%, 12.3%, and 28.0%, respectively. Optimization of mitochondrial targeting increased fluorescence 2.7 fold. During CCCP induced mitophagy, lysosomal colocalization of MagLOV2 increased from 20.75% to 89.73%, accompanied by slower MFE responses in acidic compartments. Purified MagLOV2 showed response times ranging from 8.27 s at pH 4.5 to 1.56 s at pH 9.5. Structural, spectral, and kinetic analyses suggested that protonation dependent flavin photochemistry and the surrounding protein environment contribute to this behavior. These measurements demonstrate that genetically targeted MagLOV2 can follow organelle transport and report local acidification through changes in magnetic-response kinetics.

biophysics↗

bioRxiv2026

Ultra-high-efficacy Wireless Neuromodulation Using a Conductive Ring Interface

An ab-initio physics study, supported with a COMSOL-based numerical simulation, presents a concept to enable wearable, highly targeted, and ultra-high-efficacy neuromodulation. A target site is defined by a relatively small gap of a miniaturized ring-shaped conductive interface implant. If the implant conductivity significantly exceeds the conductivity of the biological tissue, a highly localized electromotive force (emf) is induced in the ring upon application of an alternating current (AC) magnetic field. This emf generates an electric field in the gap, inducing local neural stimulation of the targeted site, e.g., translated to a nerve segment using circumferential cuffs. The wirelessly controlled autonomous stimulator implant acts as a low-resistance nearly-closed path to generate a highly localized electric field on the order of 50 V/m in the gap via application of a 1-kA/m AC magnetic field at a frequency of 10 kHz. The magnetic field is at least three orders of magnitude smaller than the magnetic field required to generate the same electric field strength via conventional transcranial magnetic stimulation. Therefore, this high-resolution, orientation-specific ring neuromodulation system could enable a highly localized, energy-efficient wearable device with significant preclinical and clinical potential.

bioengineering↗

bioRxiv2026

Engineering the effector specificity of the macrolide-sensing transcription factor MphR for the detection of sesquiterpenes

Terpenes are a large, diverse class of natural products used as flavors, fragrances, pharmaceuticals, and more, yet obtaining them often requires either large quantities of biomass or complex chemical reactions. Directed evolution of key enzymes and pathways could improve biosynthetic yields, but most sesquiterpenes lack high-throughput screening methods to search large libraries of variants. To alleviate this bottleneck, we developed the first reported genetically encoded biosensor to detect four highly valuable sesquiterpenes: valencene, nootkatone, artemisinin, and cedrol. Starting with the macrolide transcription factor biosensor MphR, we screened a panel of mutants against the four sesquiterpenes, including MphR-13 (T17N/V18I/K21L/S106L/H147L/P161L), which readily detected all four sesquiterpenes. Computational modeling showed that the mutations reduced the size of the ligand-binding pocket compared with the wild type, which positively correlated with the biosensor's fluorescence output. Additionally, swapping the fluorescence reporter improved the sensor's response 15-20-fold, yielding a generalist sesquiterpene biosensor that can be further developed and applied to the directed evolution of sesquiterpene biosynthesis.

synthetic biology↗

bioRxiv2026

Self-extending DNA recording with stochastic and programmed writing

Recording cellular histories requires new information to accumulate without erasing earlier records. Here we introduce Infinite Printer, a CRISPR recording architecture that couples self-extension to regeneration of a writable terminal context. A self-targeting prime-editing guide RNA (stpegRNA) directs repeat extension, while base editing generates diversity in situ without a separate insertion template for each character state. Earlier sequence remains upstream of the renewed terminus, linking substrate growth with record retention. Auxiliary pegRNAs specify payloads within a write-renew-write architecture. After three rounds of payload input, 4,436 distinct complete CCA-based records contained identities from all three pools. Longer AC/AT records retained shared upstream marks, and full-length sequence analysis resolved additional endpoint distinctions. Joint single-cell readout recovered 1,549 distinct records and 7,669 combinations across 12,302 transcriptome-linked cell labels from a single stpegRNA design. Spatial-transcriptomic readout recovered spacer records at 11,330 capture spots, linking record content to tissue coordinates and gene-expression profiles. Infinite Printer unites substrate renewal, sequence diversification and programmed writing with cell- and tissue-linked readout, providing a foundation for future lineage recording across large cell populations during mammalian development.

synthetic biology↗

bioRxiv2026

Nose-to-Knee Transplantation for Osteochondral Regeneration: A Proof-of-Concept Study in Sheep Employing Ceramic-Hydrogel Osteochondral Grafts

Osteochondral defects remain a major clinical challenge, frequently progressing to osteoarthritis and ultimately requiring joint replacement. Despite advances in regenerative strategies, current approaches often fail to restore the structural and functional complexity of the osteochondral unit, resulting in poor integration and limited long-term performance. Here, we developed autologous osteochondral grafts combining mechanically mature nasal chondrocyte-derived cartilage with a 3D-printed strontium-hardystonite-gahnite trabecular bone scaffold, which was tested in an osteochondral defect model in sheep. Nasal chondrocytes were encapsulated in a hyaluronan transglutaminase-alginate hydrogel generating hyaline-like cartilage with physiologically relevant compressive properties (1.0 +/- 0.2 MPa) after 6 weeks of in vitro maturation. MicroCT scans of the osteochondral unit were used to design biomimetic trabecular bone scaffolds. Controlled infiltration of the hydrogel into the trabecular pores and subsequent cartilage maturation created a mechanical interlock between cartilage and bone layer, reaching an interfacial shear strength of 201 +/- 42 kPa. The translational potential of this approach was evaluated in a proof-of-concept ovine study. The engineered cartilage remained attached to the bone scaffold for three months following implantation, but osteochondral repair was limited by implant sinking, likely caused by inadequate implant fixation in the defect, resulting in micromotion, synovial fluid infiltration, osteolysis and subsequent bone cyst formation. Together, this work demonstrates the feasibility of engineering mechanically mature autologous osteochondral grafts while identifying subchondral bone integration as the primary bottleneck limiting successful osteochondral regeneration and future clinical translation.

bioengineering↗

bioRxiv2026

Coronavirus nucleocapsid proteins differentially antagonize RIG-I-like receptor signaling to shape pathogenicity

Coronaviruses cause human infections with divergent clinical outcomes. Viral proteins likely play important roles in the severity of human diseases but are poorly defined. In this study, we discovered that viral nucleocapsid (N) proteins from various human coronaviruses have distinct abilities to antagonize human innate immune responses. N proteins from more pathogenic coronaviruses, such as SARS-CoV and SARS-CoV-2, have more potent activities specifically against human RIG-I-like receptor (RLR) activation than less pathogenic coronaviruses including OC43 and HKU1. Mechanistically, the SARS-CoV-2 N protein interacts with RLR pathway viral RNA sensors MDA5 and RIG-I and inhibits their interaction with MAVS. The CTD of SARS-CoV-2 N protein is an important region for RLR antagonism. Consequently, N proteins from more pathogenic coronaviruses promote viral replication more effectively than N proteins from less pathogenic coronaviruses. Blocking N protein function may be a promising antiviral strategy.

microbiology↗

bioRxiv2026

Defense-associated reverse transcriptases synthesize DNA in a telomerase-like manner

Defense-associated reverse transcriptases (DRTs) constitute an emerging class of bacterial defense systems that harness reverse-transcriptase activity to generate cDNA products and counter phage infection. A distinct subset of DRT10 systems evolutionarily related to telomerase generates tandem-repeat DNA to confer antiphage immunity. Here, we biochemically characterize the reverse transcription activity of DRT10 and determine cryo-EM structures of DRT10-ncRNA complex in three functional states, revealing an unanticipated 2:1 RT:ncRNA assembly mode. Two DRT10 protomers dimerize through a central pair of interlocked {beta}-hairpins and are further embraced by a single ncRNA. This architecture distinguishes DRT10 from previously characterized DRT complexes, which either lack an ncRNA component or exhibit an equivalent stoichiometry of RT:ncRNA. Strikingly, despite the nearly identical conformations of the two DRT10 protomers, cDNA product density was resolved in only one catalytic pocket, revealing different product occupancy within the dimer. Moreover, along with flanking linkers, the stem loops of the ncRNA as fixed points to supply steric barriers help define the template boundary. These findings reveal a distinctive ncRNA-mediated DRT10 dimeric assembly and establish a mechanistic framework for protein-primed reverse transcription constrained by the length of the RNA flanking linker.

biochemistry↗

bioRxiv2026

The adaptive architecture of tRNA dependencies across physiological tumor environments

The genetic code is decoded by a highly redundant transfer RNA (tRNA) repertoire, yet whether this apparent redundancy serves a functional role beyond ensuring robust translation remains unclear. Here, we establish an atlas of tRNA dependencies through large-scale CRISPRi perturbation mapping across matched in vivo tumor growth and in vitro cell culture for 16 cancer cell lines spanning seven tissue types. During in vivo tumor growth, tRNA dependencies exhibited tissue organization at isodecoder resolution and were associated with codon demand at the isoacceptor level. Strikingly, these relationships were reorganized when cells were cultured in vitro, establishing environmental sensitivity of tRNA dependency--in contrast to protein components of the decoding machinery such as aminoacyl-tRNA synthetases (aaRSs), whose dependencies were largely preserved across environments. Despite pronounced context specificity, environmental alterations in tRNA dependency were dominated by a shared response across cancer cell lines, organized across isoacceptor and isodecoder levels. By disentangling stable cellular identity from environment-associated dependency states, we resolved a shared axis of isoacceptor dependency remodeling and identified specific tRNA isoacceptors associated with promoting or restraining adaptation to the tumor environment. We further connect this functional architecture to differential translation of codon-usage-biased programs associated with proliferation and tumor-microenvironmental stress. Together, our study reveals that tRNA redundancy does not imply functional equivalence, but instead forms a structured and environmentally plastic control layer linking coding information to translational output.

genomics↗

bioRxiv2026

Neutrophils orchestrate pre-metastatic immunosuppressive bone marrow remodeling during prostate cancer metastasis

The dynamic reprogramming of the bone marrow (BM) microenvironment during metastatic progression is poorly defined. Through single-cell profiling of BM samples spanning benign conditions, high-risk bone metastatic potential localized prostate cancer (HMP) and established bone metastatic prostate cancer (PCa), we uncovered a primary tumor-to-BM signaling axis that orchestrates niche formation through neutrophil-mediated remodeling. Tumor-associated neutrophils (TANs) emerge during the HMP stage, persist throughout metastatic progression, correlate with poor patient prognosis, and exhibit direct immunosuppressive activity that contributes to a BM environment permissive to metastatic colonization. Functional and genetic studies showed that tumor-associated systemic IGF1 reprogrammed BM neutrophils through IGF1R-MAPK/ERK-KLF4 signaling, with IL-1{beta}, reactive oxygen species and nitric oxide contributing to TAN-mediated T cell suppression. Tumor-cell Igf1 perturbation, IGF1R blockade and neutrophil-specific Igf1r deletion attenuated TAN-associated BM remodeling, and reduced subsequent bone metastatic progression. In bone-metastasis models, combined IGF1R and PD-1 blockade suppressed bone-metastatic progression. Translationally, our investigator-initiated clinical trial (NCT06866548) of teprotumumab plus tislelizumab in patients with established metastatic castration-resistant prostate cancer (mCRPC) provided initial safety data and exploratory clinical observations. Together, these findings identify IGF1R-dependent neutrophil reprogramming as a functionally important component of the evolving PCa bone metastatic microenvironment and provide a rationale for therapeutic targeting of IGF1R signaling together with immune checkpoint blockade.

cancer biology↗

bioRxiv2026

Engineered CTLA-4 targeting chimeras for targeted protein degradation and immune synapse reprogramming

Despite the success of immune checkpoint blockade in cancer therapy, overcoming the immunosuppressive tumor microenvironment remains a major barrier to durable responses. Here, we introduce cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4)-targeting chimeras (cTACs), a class of synthetic molecules that harness CTLA-4-mediated trans-endocytosis (TE) for targeted protein degradation across cellular boundaries. The lead candidate, cTAC4.0, is composed of the extracellular domain of CD80 fused to an anti-programmed death-ligand 1 (PD-L1) fragment via a human immunoglobulin G4 (IgG4) heavy chain. Unlike existing protein degraders that function within cancer cells via house-keeping pathways, cTAC4.0 operates in trans between immune and tumor cells. It redirects the natural CD80-CTLA-4 internalization to capture and degrade PD-L1 from tumor cells into T cells, while simultaneously converting PD-L1-mediated suppression into CD28-driven co-stimulation, depending on the relative availability of CTLA-4 or CD28. Mechanistically, we demonstrate that cTAC4.0 promotes CTLA-4-dependent internalization and lysosomal degradation of both membrane-bound and soluble PD-L1, reduces surface PD-L1 levels on target cells, and activates T cells in a CD28-dependent manner. Functionally, cTAC4.0 enhances the cytotoxicity of both bispecific T cell engagers and T cell receptor (TCR)-engineered T cells in vitro, and when combined with these engineered T cells, it suppresses tumor growth in vivo in a glioblastoma xenograft model. By coupling immune checkpoint degradation with T cell activation in an immune synapse-enriched manner, cTACs establish a paradigm for intercellular protein degradation systems that simultaneously deplete suppressive ligands and enhance T cell function to overcome resistance in cancer immunotherapy.

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

The multiscale architecture of trait aggression: from meta-analytic convergence to multimodal characterization

Background. Trait aggression shows substantial interindividual variability, yet its neurobiological architecture across structural, functional, molecular, and macroscale organizational levels remains poorly understood. Methods. We developed a convergent multiscale framework to characterize the neurobiological architecture associated with trait aggression. We first integrated a large-sample voxel-based morphometry analysis (n = 922) with published neuroimaging studies using two complementary coordinate-based meta-analytic approaches, activation likelihood estimation and seed-based d mapping (total N = 3,671). Convergent meta-analytic findings were subsequently mapped onto functional connectivity networks, and their molecular signatures were characterized using imaging transcriptomics and neurotransmitter mapping. Finally, we examined whether the spatial organization of aggression-related networks was embedded within intrinsic multimodal brain hierarchies spanning morphometric, functional, transcriptomic, and neurochemical dimensions. Results. Across complementary analyses, trait aggression converged on a reproducible structural-functional network centered on the dorsomedial prefrontal cortex and dorsal anterior cingulate cortex. This architecture exhibited distinct molecular signatures, including gene-expression patterns enriched for synaptic functions and robust spatial associations with serotonergic, dopaminergic, and noradrenergic transporter systems. Moreover, the spatial organization of the aggression-related functional network was systematically predicted by the brain's intrinsic multimodal gradients, indicating that it is embedded within broader principles of cortical hierarchical organization. Conclusions. these findings move beyond the localization of aggression to discrete brain regions and identify a reproducible multiscale architecture linking aggression-related structural variation to functional network organization, molecular signatures, and the brain's intrinsic multimodal hierarchy. This integrative framework provides a systems-level account of the neurobiological basis of individual differences in trait aggression.

neuroscience↗

bioRxiv2026

Discrete Morse Graph Construction for High-Dimensional Transcriptomic Data

Single-cell transcriptomic atlases resolve thousands of putative brain cell types, yet common analytical workflows often rely on low-dimensional embeddings that can distort neighborhood relationships and obscure continuous variation between closely related populations. We introduce single-cell Discrete Morse Graph Construction (scDMGC), a deterministic framework based on topological data analysis and discrete Morse theory that extracts a compact graph from neighborhood structure in high-dimensional gene-expression space. Because graph edges are derived from the original k-nearest-neighbor complex, scDMGC provides an interpretable multiscale representation of transcriptomic organization without requiring a low-dimensional embedding. Local maxima represent coherent transcriptional states, saddles quantify their separation, and gradient paths capture intermediate states and continuous transitions. Applied to cortical and hippocampal datasets, scDMGC recovers established inhibitory-neuron classes and elucidates transcriptional gradients. In mouse whole-cortex data, Morse graph structure evaluates cell-type hierarchy and type robustness. Persistence across scales provides a quantitative measure of cell-type identity and discrete versus continuous relationships. Finally, in Alzheimers disease single-nucleus data, scDMGC distinguishes changes in cell-type abundance from disease-associated shifts in transcriptional state. scDMGC therefore provides a multiscale framework for quantifying whether transcriptional populations form persistent cell types, how strongly external annotations are supported by intrinsic data structure, how cell types relate to one another, and where discrete organization transitions into continuous variation.

neuroscience↗

bioRxiv2026

RNASeek: A Cross-Phyla Generative Foundation Model for Multipurpose RNA Modeling and Reinforcement Learning-Based Design

RNA plays central roles in regulating information flow and provides a versatile substrate for engineering biological functions. While large language models (LLMs) have transformed natural language processing and protein design, a general framework connecting RNA foundation models to functional sequence design remains limited. Here, we present RNASeek, a 1.6-billion-parameter generative foundation model built on a DeepSeek architecture and trained on a cross-phyla transcriptomic corpus for RNA sequence representation and generation. Natural-language tokens enable flexible conditional prediction and sequence design using a unified backbone. RNASeek captures species-specific transcript features and intron-exon boundaries in a zero-shot setting. We then fine-tune RNASeek to predict ribozyme self-cleavage activity and viral mRNA stability, revealing interpretable sequence features associated with function, including ribozyme loop flexibility and stem stability, as well as AU-rich motifs associated with mRNA stability. We use these functional predictors as reward models and apply Group Relative Policy Optimization (GRPO) to update the generation policy of RNASeek toward sequences with desired properties. GRPO-guided generation produces faster-cleaving ribozymes and stability-enhancing 3' UTRs while satisfying user-specified IUPAC constraints. Experimentally validated RNASeek-generated ribozymes achieve wild-type levels of activity, while RNASeek-generated 3' UTR sequences exceed the performance of the training data and benchmarked AI-generated 3' UTRs. Together, RNASeek establishes a unified pretrain-predict-optimize framework that connects learned RNA function to controllable de novo sequence design and provides a general strategy for engineering regulatory RNAs with desired properties.

bioengineering↗