Biology subjects

Hu, B.

Publications and source records attributed to Hu, B..

At least 19 recordsLinked to original sources

bioRxiv2026

ProtCell: a foundation model for single-cell proteomics

Single-cell proteomics provides a direct functional readout of cellular states, but targeted assays measure limited and non-identical protein panels, fragmenting information across experiments. Here, we present ProtCell, a foundation model developed using approximately 380 million cells from 98 datasets for pretraining and evaluation. ProtCell combines protein-identity and continuous-abundance embeddings in a transformer encoder and learns cell- and protein-level representations through masked abundance modeling. ProtCell first connects fragmented profiles through transferable annotation and batch correction while preserving biological organization. It then extends limited profiles through abundance reconstruction and inference of an unmeasured cytotoxic protein program enriched in natural killer cells and memory CD8$^+$ T cells. The learned protein dependencies further guide the selection of compact panels that retain information for profile reconstruction and biological classification. ProtCell thus provides a framework for connecting fragmented proteomic profiles, extending information beyond measured panels and guiding more informative measurements.

bioinformatics↗

bioRxiv2026

Crystalline silica exposure causes latent biomechanical abnormalities along the cardiopulmonary axis

Rationale Crystalline silica exposure is a major global health burden linked to chronic lung disease and cardiopulmonary morbidity, but early lung-vascular-cardiac abnormalities preceding overt functional decline are poorly defined. Objectives To determine whether a single crystalline silica exposure produces subclinical abnormalities across the integrated axis of cardiopulmonary response (lung[->]pulmonary artery[->]right ventricle) and identify associated sex-dependent physiologic patterns and molecular programs. Methods Male and female mice received a single endotracheal instillation of crystalline silica or vehicle (saline). At 28 days, assessments included voluntary wheel running, diffusion capacity, pulmonary function testing, bronchoalveolar lavage analysis, lung histology and electron microscopy, ex vivo proximal pulmonary artery biomechanics and microstructural second harmonic generation imaging, right ventricular pressure, transthoracic echocardiography, and bulk RNA sequencing of lung and right ventricle. Results Despite recovery of body weight and preserved voluntary exercise, silica-exposed mice demonstrated reduced diffusion capacity, ultrastructural thickening, and disruption of the alveolar-capillary barrier. Bronchoalveolar lavage revealed persistent inflammation with elevated macrophages, lymphocytes, neutrophils, and protein, alongside retained particulate and surfactant-rich debris (silicoproteinosis). Lung mechanics were sex-dependent; females showed a parenchymal-dominant phenotype and males showed increased airway resistance. Elevated pulmonary circulatory pressures caused increased functional stiffness and reduced distensibility of the proximal pulmonary arteries. Right ventricular free wall thickening was noted in females. Lung transcriptomics indicated persistent immune activation and extracellular matrix remodeling, while the right ventricle demonstrated metabolic reprogramming with downregulation of structural and reparative programs. Conclusions A single crystalline silica exposure induces a subclinical, multi-compartment cardiopulmonary phenotype characterized by persistent alveolar-capillary injury, pulmonary vascular stiffening, increased right ventricular afterload, sex-dependent pulmonary mechanics, and divergent organ-level transcriptomic responses.

physiology↗

bioRxiv2026

Selectivity Dilution in Three-Factor Learning: A Finite-Population Model with Potential Implications for Motor Development and Post-Stroke Motor Rehabilitation

Motor development and post-stroke motor recovery may share underlying learning mechanisms, but the early mechanical environments of the two processes differ strikingly. During early development the fetus moves within amniotic fluid, where buoyancy substantially reduces gravitational load and enables spontaneous low-resistance movement; in contrast, post-stroke motor reconstruction occurs without this fluid low-resistance phase. Whether and how the absence of a fluid low-resistance phase influences post-stroke functional recovery is the central testable question raised by the selectivity dilution framework. Three-factor learning rules use broadcast neuromodulatory signals to convert eligibility traces into lasting synaptic changes. Because the signal is broadcast, it may reinforce all eligible synapses that are active during reinforcement without obvious deviation, not only those that caused the successful outcome. This framework offers a testable mechanistic principle for early motor learning and neurorehabilitation, and may inform the design of biologically plausible, non-backpropagation learning algorithms in artificial systems.

neuroscience↗

bioRxiv2026

DeepD enables efficient genome modeling and controllable long-context sequence generation across domains of life

Genome foundation models hold great potential for advancing biological research, but modeling diverse complex genomes and generating controllable genome-scale sequences remain challenging. We present DeepD, a phylogeny-informed sparse mixture-of-experts model operating at single-nucleotide resolution over a 1-million-token context while reducing active computation. Pretrained on unselected whole-genome sequences across all domains of life, DeepD learns genic and intergenic sequence features and supports broad sequence-to-function prediction across species. DeepD performs well in human protein fitness and disease variants effect prediction, such as BRCA1 and BRCA2, with particular strength in regulatory regions, such as promoter and transcription-factor binding sites. Expert routing captures exon-intron boundaries, transposable elements, and distal cis-regulatory elements. DeepD also generates experimentally validated core promoters with specified transcriptional strengths. Finally, the Context-Anchor-Body strategy enables generation of complete yeast chromosome sequences and megabase-scale Arabidopsis segments, preserving gene structure and order while introducing variations within natural ranges. Together, DeepD is an efficient genome foundation model that unifies accurate sequence-to-function prediction with controllable genome-scale generation, advancing predictive and generative design of complex genomes.

bioinformatics↗

bioRxiv2026

Adenosylhomocysteinase drives epigenetic and metabolic reprogramming in ZFTA-RELA ependymoma

Supratentorial ependymomas (ST-EPNs) are aggressive pediatric brain tumors predominantly driven by the recurrent ZFTA-RELA fusion, an oncogenic chromatin regulator that establishes an aberrant transcriptional program. Although metabolic pathways modulate epigenetic regulation by controlling the availability of metabolic cofactors, how ZFTA-RELA coordinates metabolic reprogramming with epigenetic regulation is not well understood. Here, we show that ZFTA-RELA-positive ependymomas exhibit enhanced methionine dependency, elevated adenosyl homocysteinase (AHCY) expression, and epigenetically control aspartate metabolism for tumor growth. Mechanistically, ZFTA-RELA upregulates AHCY, increasing cellular methylation potential which promotes H3K4me3-dependent expression of CAD, the first enzyme complex of de novo pyrimidine biosynthesis, while simultaneously enforcing H3K27me3 mediated repression of ASNS, which encodes asparagine synthetase. Through this coordinated epigenetic program, aspartate utilization is directed toward pyrimidine biosynthesis while endogenous asparagine production is limited. Restricting S-adenosylmethionine availability or genetically or pharmacologically inhibiting AHCY reverses these epigenetic states, leading to CAD downregulation, ASNS upregulation, impaired nucleotide synthesis, and suppression of tumor growth in vitro and in vivo. Similarly, EZH2 inhibition also restores ASNS expression and suppresses tumor proliferation. These findings suggest that ZFTA-RELA EPN maintain ASNS-low phenotype which conditionally make them dependent on extracellular asparagine. We further show that enhanced de novo pyrimidine synthesis not only supports nucleotide production but also provides UDP-sugars required for N-glycosylation and stabilization of the asparagine transporter SLC1A5, thereby promoting extracellular asparagine uptake and tumor proliferation. Collectively, our findings identify AHCY as a critical metabolic regulator of epigenetic and metabolic reprogramming that drives pyrimidine synthesis and establishes extracellular asparagine dependency, revealing actionable metabolic vulnerabilities in ZFTA-RELA-positive ST-EPNs.

cancer biology↗

bioRxiv2026

Adenosine refines the balance of spermatogonial proliferation and differentiation

The balance between spermatogonial proliferation and differentiation is essential for sustaining male fertility, yet the contribution of cellular metabolism remains poorly understood. Here, we showed that adenosine levels increased significantly during spermatogonial differentiation and promoted the formation of differentiating spermatogonia. Using a conditional knockout mouse model in which Ahcy, a gene encoding an enzyme in adenosine synthesis, is deleted from spermatogonia, we found that disruption of adenosine generation blocked spermatogonial differentiation, resulting in a transient delay in spermatogenesis. We further demonstrated that adenosine exerted its effects partially through ADORA2A- and ADORA2B-mediated signaling. Downregulating these receptor activities by specific inhibitors in spermatogonial culture or by genetic ablation in mice led to increased spermatogonial proliferation. Taken together, our study identified adenosine as a metabolic signal in refining the balance between spermatogonial proliferation and differentiation, linking cellular metabolism to critical developmental timing and stem cell fate decisions.

developmental biology↗

bioRxiv2026

EnsPlex: Integrative Protein Complex Prediction through Multi-source Complementary Structural Sampling and Topology-Aware Candidate Selection

Protein complex structure prediction depends on both the breadth of candidate coverage and the ability to select accurate models within a limited output budget. Most existing approaches emphasize only one stage. End-to-end models and molecular docking workflows generate candidate structures, whereas quality-assessment models primarily rerank a predefined candidate pool. A complete strategy must account for complementary sampling across generators, differences among scoring scales and the allocation of final candidate quotas. Here, EnsPlex is presented as a multi-source framework that couples structural sampling with candidate selection for protein-protein interaction complex prediction. EnsPlex combines conformation-expanded docking, AlphaFold-Multimer, AlphaFold3 and Boltz-1 to expand the sampled conformational space. Conformation-expanded docking comprises monomer conformational expansion followed by flexible HADDOCK docking. FACET is trained to predict candidate quality using DockQ and its component metrics as supervision. Across 102 antigen-antibody systems, EnsPlex achieved up to a 27.8% relative improvement in target success rate over AlphaFold3 with its built-in ranking under matched output budgets. FACET also improved within-source ranking in the internal candidate pools and in homology-filtered external data. These findings support the utility of EnsPlex for structure prediction in the evaluated antigen-antibody systems.

bioinformatics↗

bioRxiv2026

SurfGraphPro: Integrating Protein Language Models with Geometric Deep Learning On Coarse Protein Surfaces for Binding Site Prediction

Accurate and fast prediction of protein binding sites remains essential for understanding molecular interactions and facilitating protein engineering. Here, we present SurfGraphPro, a geometric deep learning approach that bridges protein language models with surface-based structural representations for a rapid binding interface prediction. Our method operates on coarse-graphed triangulated protein surfaces, utilizing solvent-excluded surface meshes downsampled into amino acid residue centered patches. Unlike existing approaches that rely on extensive physicochemical feature engineering, we leverage evolutionary information directly from protein language model embeddings, eliminating the need for computationally expensive multiple sequence alignments and hand-crafted features. This integration achieves on average $\sim18-28 \times$ speedup for proteins of $\sim$100 to a few 1000s of amino acids over current state-of-the-art surface-based model while maintaining comparable accuracy on diverse binding interfaces, including challenging antibody-antigen complexes. To our knowledge, this represents the first approach to integrate protein language model embeddings with coarse geometric surface representations for binding site prediction, demonstrating that learned evolutionary features coupled with geometric transformers can replace traditional feature engineering without sacrificing performance.

bioinformatics↗

bioRxiv2026

Functional reorganization of large-scale brain systems underlies emotion regulation maturation

Emotion regulation via cognitive reappraisal is essential for socioemotional development and mental health, yet the organizational logic of its supporting brain systems across youth remains unclear. Using a large multisite cohort, we demonstrate that maturation of reappraisal capacity reflects reorganization of a distributed architecture rather than progressive strengthening of prefrontal control. Reappraisal robustly reduced negative affect, with non-linear gains concentrated between 6 and 10.5 years. Bayesian systems identification resolved this architecture into four functionally distinct systems-reappraisal-specific, common-appraisal, modifiable-emotion, and non-modifiable-emotion components-spanning prefrontal-control, salience, default-mode and dorsal-attention networks. Age-related variation in anterior temporal cortex, emerged as essential hub, statistically accounted for developmental gains in reappraisal success. Multivariate modelling together with intersubject analyses further identified a cross-site-generalizable whole-brain reappraisal signature and revealed an age-dependent shift in brain-behaviour coupling, with neurobehavioural convergence increasing with age. These findings establish distributed network reorganization as a core mechanism of emotion regulation development and provide mechanistic targets for stage-matched interventions.

neuroscience↗

bioRxiv2026

ACAN regulates VSMC mitochondrial homeostasis through the YAP1/TAZ pathway and contributes to the onset and progression of aortic dissection

Objective: Aortic dissection (AD) is a serious, life-threatening cardiovascular crisis. While vascular proteoglycans normally buffer hemodynamic stress, their aberrant accumulation in the AD aorta predisposes to rupture. Aggrecan (ACAN) is among the most frequently elevated proteoglycans in this setting, yet its pathogenic relationship to AD remains undefined. This study aimed to investigate the role of ACAN in AD and identify potential therapeutic targets. Approach and Results: We employed human tissues, animal models, and cell cultures. Western blotting and immunohistochemistry (IHC) were performed on aortic samples from AD patients and controls to assess contractile markers, MMPs, ACAN, and SOX9. In vivo, AD was induced in mice using {beta}-aminoisobutyric acid (BAPN) and Ang II, with AAV-mediated ACAN knockdown in VSMCs. Histological staining confirmed model validity and enabled assessment of AD incidence, mortality, and protein expression changes. In vitro, human aortic vascular smooth muscle cells (HAVSMCs) were stimulated with Ang II, and SOX9 and ACAN were silenced via siRNA to evaluate the functional impact of ACAN downregulation. In human AD aortas, elevated ACAN and its upstream transcription factor SOX9 drove VSMC phenotypic switching and ECM degradation. In mice, ACAN knockdown corrected AD-induced medial structural disruption, elastin fragmentation, and collagen deposition, markedly reducing incidence and mortality while curbing excessive mitochondrial fission and rescuing functional integrity. ACAN silencing attenuated Ang II-induced mitochondrial damage in HAVSMCs via the RNA-seq-identified YAP1/TAZ pathway, and Verteporfin restored homeostasis. Conclusion: ACAN is aberrantly expressed in AD and murine aortas. ACAN downregulation delays disease progression by preserving mitochondrial homeostasis and function via the YAP1/TAZ pathway.

pharmacology and toxicology↗

bioRxiv2026

Cell painting with amphiphile-protein conjugate vaccines drives B cell activation and germinal center priming to enhance humoral immunity

Most licensed vaccines against infectious diseases mediate protection by priming B cells and humoral immune responses, where efficacy largely relies on activation of germinal centers (GCs) in lymphoid tissues that serve as the training ground for a high-affinity antibody response. While subunit vaccines offer safety advantages, they tend to be poorly immunogenic with weak and short-lived antibody responses, presenting a need for engineering strategies that enhance immunogenicity. Amphiphile-protein vaccines, consisting of protein antigens modified with an albumin-binding lipid tail via a polyethylene glycol (PEG) linker, hitchhike on albumin following subcutaneous injection to enhance lymphatic trafficking and antigen-specific immune activation. Amph-vaccines also demonstrate an ability to insert their lipid tail into cell membranes, effectively 'painting' cells with multivalent antigen. We hypothesized that cell painting may play a role in driving humoral immunity by sustaining antigen persistence in lymph nodes and generating multivalent antigen presentation for B cell activation. Here, we investigated how amph-vaccine molecular properties (antigen MW and PEG linker length) influenced conformational behaviors (micelle formation, albumin hitchhiking, and membrane insertion) along with the resulting humoral immune response. Using HIV env proteins as model antigens (eOD monomer and MD39 trimer), we investigated cell membrane insertion and B cell activation via calcium flux assay in vitro, followed by lymphatic trafficking and immunogenicity studies in vivo. Our results demonstrate that amphiphile conjugation is an effective strategy for enhancing humoral immunogenicity of both monomer and trimer protein antigens, where cell painting plays a significant role in driving B cell, GC, and humoral immune activation.

bioengineering↗

bioRxiv2026

CryoFlex characterizes structural motion between conformations directly from cryo-EM density maps

Heterogeneous cryo-EM reconstruction methods can resolve multiple conformational states, but understanding their functional implications often requires determining which regions move between states and quantifying their displacement magnitudes. Comparing atomic models can quantify motion, but flexible regions are often incompletely modeled. Direct map comparison is limited by variable map quality and ambiguous correspondence between displaced density features. Here, we present CryoFlex, a method that estimates the direction and magnitude of structural motion directly from reconstructed density maps of selected states. The resulting motion analysis supports functional interpretation and guides downstream structure processing. CryoFlex localized and quantified structural motion across diverse map pairs, including those with incomplete atomic-model coverage or weak local density. Against C; displacements, CryoFlex achieved an endpoint error of about 1 angstrom. In separate benchmarks, it outperformed registration baselines on clean and noisy inputs. Thus, CryoFlex complements heterogeneous reconstruction workflows with direct quantification of structural motion between reconstructed states.

bioinformatics↗

bioRxiv2026

Evolutionary Origins and Synthetic Reconstruction of Gingerols

How new enzymatic activities become integrated with existing metabolic networks to generate lineage-restricted plant chemistry remains a central question in metabolic evolution, exemplified by gingerol biosynthesis in Zingiberaceae, whose complete enzymatic basis has remained elusive. Here, we resolve this problem by elucidating the evolutionary origins and the full biosynthetic pathway of gingerol through single-cell transcriptomics, functional screening, structural biology, comparative genomics and pathway reconstruction. A coordinated metabolic program operates in ginger oil cells, thereby guiding the identification of diketide-CoA synthases, gingerol synthases (GS) and gingerol reductases that together constitute the complete pathway. GS remodels the conserved type III polyketide synthase scaffold to organize a fatty acyl-CoA and a phenylpropanoid-derived intermediate for gingerol formation. Furthermore, GS arose through duplication and limited active-site remodeling of a curcumin synthase-like enzyme, accompanied by coordinated recruitment of precursor-supply and reduction pathways. Translating these principles into yeast achieves the first de novo microbial production of 6-gingerol from glucose at 12.4 mg/L. These findings reveal the evolutionary logic by which plants recruit and coordinate novel enzymatic activities to establish lineage-restricted metabolic program, providing a framework for understanding and engineering specialized metabolism across the plant kingdom.

synthetic biology↗

bioRxiv2026

iPSC-Derived Microglia-like Cells Exhibit Protocol-Dependent Transcriptomic Features and Robust Phagocytosis of Glioma Cells

Microglia are the brain-resident macrophages and key regulators of the brain tumor microenvironment. Although induced pluripotent stem cell-derived microglia (iMG) provide a valuable model for studying human microglial, systematic comparisons of differentiation protocols are limited, and their utility for modeling microglia-tumor cell interactions remains underexplored. Here, we analyzed 54 public RNA-seq datasets representing 22 iMG differentiation protocols, including embryoid body (EB)-based, two-dimensional (2D), transcription factor-induced, and coculture-based approaches. Most iMG closely resembled primary human microglia, although substantial protocol-dependent differences were observed. iMG generated using EB-based protocols showed higher TMEM119 expression, whereas those generated using 2D-based protocols showed higher P2RY12 expression. A widely adopted EB-based protocol showed the highest phagocytosis gene signature. Using this protocol, we generated iMG that efficiently phagocytosed patient-derived glioma stem-like cells and upregulated inflammatory and immunoregulatory genes following phagocytosis. These findings provide a transcriptomic benchmark for current iMG models and support their use in investigating microglia-glioma interactions. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=105 SRC="FIGDIR/small/739939v1_ufig1.gif" ALT="Figure 1"> View larger version (49K): org.highwire.dtl.DTLVardef@893ebforg.highwire.dtl.DTLVardef@769c08org.highwire.dtl.DTLVardef@fe06ccorg.highwire.dtl.DTLVardef@1705d70_HPS_FORMAT_FIGEXP M_FIG C_FIG

cancer biology↗

bioRxiv2026

Deep anatomical and ultrastructural classification of neurons in the zebrafish olfactory bulb

Neuronal circuits in the olfactory bulb (OB) perform computations fundamental to pattern classification including a decorrelation and normalization of odor-evoked activity. These computations are mediated by diverse interneurons but a comprehensive picture of interneuron types and their microcircuit organization is lacking. We provide a deep anatomical classification of neuron types and their synaptic connectivity in the OB of adult zebrafish, a well-established model to analyze olfactory computations. We reconstructed 459 neurons in an image volume acquired by serial block face scanning electron microscopy and defined 13 neuron classes based on morphological and ultrastructural features. These comprised two classes of projection neurons and 11 interneuron classes, some of which were further separated into subclasses. Ultrastructural information including spine shape, variations in neurite diameter and synaptic arrangements contributed significantly to the distinction of cell types. As in other species, reciprocal synaptic connections were abundant. Targeted synapse annotation revealed systematic connectivity between projection neurons and interneurons. These included microcircuit motifs combining reciprocal and unidirectional connectivity that provide possible structural substrates for gain control and lateral inhibition. The results provide detailed insights into the structural organization of the OB and an anatomical foundation for physiological and computational studies of information processing in olfaction.

neuroscience↗

bioRxiv2026

Cerebrovascular Imaging-to-Graph Reconstruction for Individualized Digital Twin Brains

The development of digital twins in medicine, i.e., virtual replicas of human organs, offers a promising path toward precision medicine by enabling interpretable, mechanistic, and actionable insights. In the brain, cerebrovascular twins support individualized modeling of hemodynamics and bio-transport, with broad applications. A major bottleneck, however, is the lack of robust methods to transform in vivo cerebrovascular images into simulation-ready cerebrovascular meshes or graphs. Here, we present CerebroVascular Imaging to Graph reconstruction (CVIG), a robust and multiscale framework for reconstructing whole brain cerebrovascular graphs from in vivo cerebrovascular images. CVIG integrates vessel vectorization, with tolerance to discontinuity in vessel structures, using a topology-guided assembly of vessel trees to generate cerebrovascular graphs from medical images. We demonstrate the ability of CVIG to generate vascular graphs with improved vascular coverage and topological correctness, the capability essential for high fidelity brain biophysical simulations. This work establishes a vascular graph framework for individualized modeling and analysis, providing a key foundation for digital twins of the human brain.

bioengineering↗

bioRxiv2026

HoloCell: A Generative Foundation Model for Holistic Cellular Modeling

Single-cell multi-omics technologies have recently advanced to enable the profiling of epigenomic, transcriptomic, and proteomic layers within individual cells, offering new opportunities to characterize cellular states as integrated biological systems. However, developing a unified framework that can seamlessly integrate diverse omics modalities and remain robust to heterogeneous modality missingness remains challenging. Existing methods are often designed for specific modalities or modality pairs, relying on dataset-specific training or paired measurements. Here we present HoloCell, to our knowledge the first generative foundation model for joint representation learning and generative modeling across all three major single-cell omics modalities, i.e., epigenomics, transcriptomics, and proteomics. HoloCell contains over 860 million parameters and is pretrained on the Human-Multi-Omics-Corpus, which comprises approximately 468 million single-cell profiles across these three omics layers, corresponding to over 425 billion tokens. HoloCell introduces a a simple yet biologically motivated hierarchical tokenization strategy that encodes cis-regulatory elements, genes, and proteins as structured tokens within a shared modeling framework. We evaluated HoloCell across single-omics representation learning, paired multi-omics integration, unpaired multi-omics alignment, and cross-modal generation via iterative diffusion and remasking, demonstrating its superior performance and flexibility across diverse omics tasks. From a representation perspective, HoloCell provides a unified digital mapping of cellular states across multiple omics layers, capturing cell heterogeneity as an integrated system. From a generation perspective, its iterative diffusion and remasking frame-work permits flexible generation orders beyond fixed left-to-right causality, enabling in silico simulation of multi-omics information flow. Together, these capabilities position HoloCell as a versatile foundation model toward the emerging concept of a virtual cell, offering both systematic characterization and generative simulation of cellular systems within a unified framework.

bioinformatics↗

bioRxiv2026

Temporal-deviation-driven community detection uncovers early-warning signals for critical transitions in complex diseases

Early detection of critical transitions in complex diseases is crucial for timely clinical intervention. However, as patients often provide only a single snapshot, identifying sample-specific early-warning signals (EWS) from a dynamical evolution perspective remains challenging, coupled with high-dimensional noise amplification. Here, we present TD-COM, a framework for detecting personalized EWS of critical transitions via single-sample community detection. By constructing a temporal perturbation map STDN, TD-COM captures latent dynamical perturbations inferred from static individual profiles. Synergizing these temporaldeviation signals with static topological features, TD-COM implements a multilevel node filtering strategy during community detection, effectively suppressing single-sample noise. Validated on hour-scale, multi-year, and multi-decade transcriptomic data, TD-COM robustly detects critical states preceding clinical deterioration and uncovers their underlying molecular mechanisms. Comparative experiments demonstrate that TD-COM outperforms existing methods in accuracy and topological robustness. Thus, TD-COM provides a generalizable framework for personalized early warning of complex diseases, particularly when longitudinal sampling is infeasible.

bioinformatics↗