Biology subjects

Xu, X.

Publications and source records attributed to Xu, X..

At least 217 records · Page 12Linked to original sources

bioRxiv2025

Spatial transcriptomics elucidates the central role of cerebral vasculatures in the progression of Japanese encephalitis

The cellular and molecular mechanisms underlying the neuropathology of Japanese encephalitis virus (JEV) remain obscure. Herein, we designed Stereo-seq chips to simultaneously capture the in situ transcriptomes of both the host and JEV, constructing a comprehensive spatiotemporal pathological landscape for Japanese encephalitis (JE). This study reveals the central role of the vascular system in JE pathogenesis, particularly the meninges, which displayed the strongest signal of inflammation and cell death in the JEV-infected brain. The activation of the Ackr1+ endothelial cells, disruption of the blood-brain barrier, the migration of JEV-infected monocytes, secretion of immune factors by the infected cells, and the occurrence of pyroptosis and necroptosis form a positive feedback loop, resulting in an increasing influx of immune cells and tissue damage. JEV infection leads to neurological impairments, which may be attributed to the interaction between viral proteins and host cellular factors such as Stat1, Stat3, Nfkb1, and Sp1. As the vascular system serves as a central receiver and amplifier of inflammatory signals, regulating inflammation within the vascular system is essential in mitigating JE progression.

immunology↗

bioRxiv2025

Improved structural modelling of antibodies and their complexes with clustered diffusion ensembles

MotivationGaining structural insights into antibody-antigen complexes is crucial for understanding antigen recognition mechanisms and advancing therapeutic antibody design. However, accurate prediction of the structure of highly variable complementarity-determining region 3 on the antibody heavy chain (CDR-H3 loop) remains a significant challenge due to its increased length and conformational variability. While AlphaFold2-multimer (AF2) has made substantial progress in protein structure prediction, its application on antibodies and antibody-antigen complexes is limited by the weak evolutionary signals in the CDR region and the lack of structural diversity in its output. ResultsTo address these limitations, we propose a workflow that combines AlphaFlow to generate ensembles of potential loop conformations with integrative modelling of antibody-antigen complexes with HADDOCK. Improving the structural diversity of the H3 loop increases the success rate of subsequent docking tasks. Our analysis shows that while AF2 generally predicts accurate antibody structures, it struggles with the H3 loop. In cases where AF2 mispredicts the loop, we leverage AlphaFlow to generate ensembles of loop conformations via diffusion-based sampling, followed by clustering to produce a structurally diverse set of models. We demonstrate that these ensembles significantly improve antibody-antigen docking performance compared to the standard AF2 ensembles. Availability and implementationThe input datasets and codes involved in this research are available at https://github.com/haddocking/alphaflow-antibodies. All the resulting modelling data are available from Zenodo (https://zenodo.org/records/14906314).

bioinformatics↗

bioRxiv2025

Effects of Inflammatory Factors on the Migration and Repair of Intestinal Mucosal Cells at Pancreaticojejunostomy Anastomosis: An Experimental Study in Beagle Dogs

ObjectiveTo investigate the effects of inflammatory factors on the migration and repair of intestinal mucosal cells at the site of pancreaticojejunostomy (PJ) anastomosis and to observe the safety and feasibility of PJ with mucosal-priority healing (PM-PJ). Materials and methodsThirty-six Beagle dogs were randomly divided into two groups: the PM-PJ group (n=18) and the control group (n=18), which included classic end-to-side invagination pancreaticojejunostomy (IPJ). Six Beagles were sacrificed on the 7th, 14th, and 21st days after surgery to obtain pancreaticojejunostomy anastomosis tissues. The primary outcomes were postoperative mortality, morbidity, and pathological changes in the anastomosis. The secondary outcomes were the expression levels of the cytokines TGF-{beta}1, IL-10, TNF-, IL-6, collagen, and -SMA. ResultsAll surgeries were successfully performed. In the study group, the incidences of anastomotic leakage and pancreatitis were 0% and 11.1% (2/18), respectively, whereas in the control group, the incidences were 5.6% (1/18) and 16.7% (3/18), respectively. The levels of serum TGF-{beta}1, IL-10, TNF-, and IL-6 were increased at all time points after surgery in both groups, and these increases were more significant at certain time points in the study group. Pathological observation revealed that the degree of anastomotic healing in the study group was greater than that in the control group at all time points. The protein expression levels of collagen I/III and -SMA in the anastomoses of both groups changed over time, and the changes were more significant at certain time points in the study group. ConclusionInflammatory factors play a positive role in the early stage of PJ anastomosis healing by stimulating collagen secretion and synthesis. PM-PJ is a safe and simple surgery with good anastomotic healing and a low incidence of complications.

animal behavior and cognition↗

bioRxiv2025

Single-nucleus multiomics of murine gonadal cells reveals transcriptional regulatory network underlying supporting lineage differentiation

Sex determination of mammalian gonads hinges upon sex-specific differentiation of gonadal supporting cells: Sertoli cells in the testis and granulosa cells in the ovary. To gain insights into how supporting cells acquire their identities, we performed joint single-nucleus transcriptomics and chromatin accessibility assays on murine gonadal cells across sex determination. By contrasting sex-specific gene expression and corresponding chromatin accessibility among progenitor and differentiated cells, we discovered that sex-specific chromatin regions in supporting cells are established shortly after sex determination, accompanied by the acquisition of active histone marks. The presence of potential transcription factor binding motifs in the open chromatin regions revealed regulatory networks underlying ovary-enriched factors LEF1 and MSX1, which promote granulosa fate by inducing granulosa-specific genes such as Foxl2 and Fst. Our results not only identify the gene regulatory framework underlying supporting cell sex differentiation, but also provide invaluable resources for the field. TeaserSex-specific changes in chromatin and gene expression underly the divergent developmental fates of the bipotential gonadal primordium.

developmental biology↗

bioRxiv2025

In vitro cellular phenotypes of cortical neurons from R255X MECP2 knock-in mice are improved by either expression of wildtype MeCP2 or read-through with G418

Approximately 60% of individuals with Rett syndrome (RTT) carry a nonsense variant in the MECP2 gene; thus, there is an unmet need to identify novel nonsense suppression compound(s) that can restore full length MeCP2 protein levels and function. Here, we characterized neuronal phenotypes in cultured cortical neurons from newborn knock-in mice harboring the MECP2 R255X variant. After 2 weeks in vitro, R255X mutant neurons showed smaller cell bodies, shorter dendrites, fewer dendritic branches, and a lower density of excitatory synapses when compared to wildtype (WT) neurons. Transduction of AAV9-MeCP2-GFP in R255X mutant neurons made these cellular phenotypes similar to those in WT neurons, including soma size, dendritic length and branching, and excitatory synapse density. As proof of principle for the potential clinical use of read-through compounds, cultured R255X mutant neurons treated with the aminoglycoside G418 for 72h in vitro showed cell body size and excitatory synapse density similar to WT neurons. We expect these combined approaches will identify effective compounds to suppress translation termination at a premature termination codon, which can be moved to further preclinical functional and behavioral studies in R255X MECP2 knock-in mice. Summary StatementExpression of wildtype MECP2 or treatment with G418 in vitro restored cell body size, dendritic length, and dendritic spine density in cortical neurons from R255X MECP2 knock-in mice to levels comparable to wildtype neurons.

neuroscience↗

bioRxiv2025

Spatial and single-cell expression analyses reveal complex expression domains in early wheat spike development

Wheat is important for global food security and understanding the molecular mechanisms driving spike and spikelet development can inform the development of more productive varieties. In this study, we integrated single-molecule fluorescence in situ hybridization (smFISH) and single-cell RNA sequencing (scRNA-seq) to generate an atlas of cell clusters and expression domains during the early stages of wheat spike development. We characterized spatiotemporal expression of 99 genes by smFISH in 48,225 cells at the early transition, late double ridge, and floret primordia stages. These cells were grouped into 21 different expression domains, including four in the basal region of the developing spikelets and three different meristematic regions, which were consistent across spikelets and sections. Using induced mutants, we revealed functional roles associated with the specific expression patterns of LFY in intercalary meristems, SPL14 in inflorescence meristems, and FZP in glume axillae. Complementary scRNA-seq profiling of 26,009 cells from W2.5 and W3.5 stages identified 23 distinct cell clusters. We used the scRNA-seq information to impute the expression of 74,464 genes into the spatially anchored smFISH-labelled cells and generated a public website to visualize them. We used experimental and imputed expression profiles, together with co-expression studies and correlation matrices, to annotate the scRNA-seq clusters. From co-expression analyses, we uncovered genes associated with boundary genes TCP24 and FZP, as well as the meristematic genes AGL6 and ULT1. The smFISH and scRNA-seq studies provided complementary tools for dissecting gene networks that regulate spike development and identifying new co-expressed genes for functional characterization.

plant biology↗

bioRxiv2025

Themis dominates T cell exhaustion by regulation of TCR and PD-1 signaling

T cell exhaustion is important to protect the host from immunopathology during chronic viral infection, but it also impairs T cell anti-tumor immunity1-5. A fundamental unresolved question is whether and how T cell exhaustion is determined at the onset of TCR signaling6-8. Here we report an unexpected role of Themis, a TCR-proximal signaling molecule9, in T cell exhaustion. Chronic viral infection in mice usually leads to T cell exhaustion and survival of the host. Surprisingly, Themis T-cell conditional knockout mice died from severe CD8+-dependent lung immunopathology in chronic viral infection, showing Themis importance in establishing T cell exhaustion. We found that Themis-deficient CD8+ T cells were hyperactivated at the single-cell level - producing more TNF and IFN{gamma} compared to wild-type counterparts - but defective in population-level expansion. Moreover, TCF-1 and TOX expression were inhibited in Themis-deficient CD8+ T cells, thereby impairing differentiation of exhausted T cell precursors (T-pex) and maintenance of terminally exhausted T cells (T-ex), respectively. Mechanistically, Themis initially promotes TCR signaling to induce PD-1 expression and subsequently mediates PD-1 signaling. In the latter, Themis binds to PD-1 and promotes PD-1 phosphorylation and its recruitment of SHP2, thereby acting as a negative regulator to inhibit T cell effector functions. Without Themis, the orderly regulation of TCR and PD-1 signaling, and therefore exhaustion, is disrupted. Thus, our results unequivocally demonstrate that Themis-mediated early TCR signaling plays a decisive role in T cell exhaustion and provide a novel mechanism of PD-1 signaling through Themis.

immunology↗

bioRxiv2025

Predicting Mutation-Induced Relative Protein-Ligand Binding Affinity Changes via Conformational Sampling and Diversity Integration with Subsampled Alphafold2 in Few-Shot Learning

Predicting the impact of mutations on protein-ligand binding affinity is crucial in drug discovery, particularly in addressing drug resistance and repurposing existing drugs. Current structure-based methods are constrained by their dependence on known protein-ligand complex structures. The absence of structural data for mutated proteins, along with the potential of mutations to modify protein conformational states, brings challenges for reliable predictions. The downstream application of conformational sampling of Alphafold, such as the prediction of binding free energy changes induced by mutation, has not been fully explored. To tackle the challenge, we propose a few-shot learning approach that integrates AlphaFold2 subsampling, ensemble docking, and Siamese learning to predict the relative binding affinity changes induced by mutations. We validate our approach for predicting binding affinity changes in 31 variants of the mutated Abelson tyrosine kinase (ABL). The conformation ensembles for each ABL variant are generated using AlphaFold2 subsampling. The most likely conformational states are selected by Gaussian fitting to construct the pairwise structural inputs and compute relative binding energies. Then the relative binding affinities are derived by averaging the values predicted by the Siamese learning model across the constructed conformational pairs. By benchmarking on the Tyrosine kinase inhibitors (TKI) dataset and the direct binding affinity using the refined set of PDBbind database, our proposed approach achieves an increased Spearman correlation coefficient for five out of six TKI molecules across 31 mutants when estimating relative binding affinities. Although the proposed method is validated on the ABL, it can be transferred and applied to other drug-target interaction predictions when the conformational flexibility of proteins needs to be considered. We anticipate the approach will be valuable for predicting binding energy changes induced by mutations in proteins or ligands, facilitating drug discovery efforts.

bioinformatics↗

bioRxiv2025

Diverse mechanisms of DDX3Y suppression by DDX3X

The DEAD-box RNA helicase DDX3X has important roles in development and disease. Loss of DDX3X during developmental and pathological processes such as tumorigenesis can lead to compensatory upregulation of the close paralog DDX3Y in males, which may underlie the sexual dimorphism displayed by some DDX3X-associated diseases. However, how DDX3X cross-regulates DDX3Y remains largely unknown. Here, we investigated the regulation of DDX3Y by DDX3X in two male-derived human cancer cell lines, HCT116 and U87MG. Depletion of DDX3X in HCT116 cells results in moderately increased DDX3Y mRNA and protein, in part due to stabilization of DDX3Y transcripts. Conversely, reduction of DDX3X in U87MG cells markedly upregulates DDX3Y protein without affecting its mRNA, mainly by enhancing DDX3Y protein stability. We further show that DDX3X physically interacts with DDX3Y. DDX3Y is much less stable than DDX3X in U87MG cells, and substitution of two lysine residues in DDX3Y with the corresponding arginine in DDX3X stabilizes DDX3Y. Thus, the compensatory upregulation of DDX3Y following DDX3X loss can occur at either transcript or protein level, suggesting complex and cell type-specific cross-regulation between these X- and Y-linked paralogs to keep the total DDX3 dosage in check.

biochemistry↗

bioRxiv2025

dandelionR: Single-cell immune repertoire trajectory analysis in R

Integration of single-cell RNA-sequencing (scRNA-seq) and adaptive immune receptor (AIR) sequencing (scVDJ-seq) is extremely powerful in studying lymphocyte development. A python-based package, Dandelion, introduced VDJ-feature space method, which addresses the challenge in integration of single-cell AIR data with gene expression data and enhanced the trajectory analysis results. However, no R-based equivalent or similar methods currently exist. To fill this gap, we present dandelionR, an R implementation of Dandelions trajectory analysis workflow, bringing the VDJ feature space construction and trajectory analysis using diffusion map and absorbing Markov chain to R, offering a new option for scRNA-seq and scVDJ-seq analysis to R language users.

bioinformatics↗

bioRxiv2025

Leopard Cats Occupied Human Settlements in China for 3,500 years before the Arrival of Domestic cats in 600-900 CE around the Tang Dynasty

The earliest cats in human settlements in China were not domestic cats (Felis catus), but native leopard cats (Prionailurus bengalensis). To trace when and how domestic cats arrived in East Asia, we analyzed 22 feline bones from 14 sites across China spanning 5,000 years. Nuclear and mitochondrial genomes revealed that leopard cats began occupying anthropogenic scenes around 5,400 years ago and last appeared in 150 CE. Following several centuries gap of archeological feline remains, the first known domestic cat (706-883 CE) in China was identified in Shaanxi during the Tang Dynasty. Genomic analysis suggested a white or partially white coat and a link to a contemporaneous domestic cat from Kazakhstan, indicating a likely dispersal route via the Silk Road. The two felids once independently occupied ancient anthropogenic environments in China but followed divergent paths and reached different destinations in human-animal interactions.

evolutionary biology↗

bioRxiv2025

Intense oocyte competition builds the female reproductive reserve in mice

During the process of ovariogenesis, more than two-thirds of germ cells are sacrificed to enhance the quality of the remaining oocytes. But the detail process of selection is not completely understood in mammals. Here, by creating a high resolution 4- dimensional ovariogenesis imaging system, we recorded the progress of oocyte fate determination in live mouse ovaries, and discovered an oocyte competition mechanism that determines the survival of oocytes. We found that dominant oocytes catch and absorb the cell debris from sacrificed oocytes to enrich their cytoplasm for survival. Single-cell sequencing demonstrated that this competition follows a classic autophagy- driven model. Blocking competition by suppressing autophagy resulted in an enlarged pool of survived oocytes. However, these oocytes lacking competition were unable to fully develop and contribute to fertility. Our study suggests that mammals have evolved a unique intensive competition system to enhance oocyte quality, necessary to support their long reproductive lifespan.

developmental biology↗

bioRxiv2025

Single-Cell Genomics Elucidates Molecular Variations and Regulatory Mechanisms in Circulating Immune Cells

The human peripheral blood displays diverse molecular characteristics across populations, understanding the drivers and underlying mechanisms of which remains challenging. Here, we introduce the Chinese Immune Multi-Omics Atlas (CIMA), elucidating sex-, age-, and genetic-related molecular variations by analyzing multi-omics data from 428 adults with over 10 million immune cells. CIMA generated an enhancer-driven gene regulatory network, identifying 237 high-quality regulons and revealing cell type-specific regulatory mechanisms. Additionally, 11,521 lead cis-expression quantitative trait loci (eQTLs) and 46,339 chromatin accessibility QTLs (caQTLs) were identified at cell type level. CIMA also uncovered pleiotropic associations among immune-related disease risk loci, eQTLs, and caQTLs in a cell type-specific manner. Lastly, a novel cell language model, CIMA-CLM, was developed to predict chromatin accessibility and noncoding variant effects using chromatin sequences and gene expressions. This work represents a population-scale multi-omics resource of human immune cells, providing a valuable reference for future investigation of immune-related diseases.

immunology↗

bioRxiv2025

RegFormer: A Single-Cell Foundation Model Powered by Gene Regulatory Hierarchies

Single-cell RNA sequencing (scRNA-seq) enables high-resolution profiling of cellular diversity, but current computational models often fail to incorporate regulatory priors, handle data sparsity, or efficiently process long gene sequences. Here, we present RegFormer, a foundation model that integrates gene regulatory networks (GRNs) with Mamba-based state-space modeling, overcoming the scalability and context-length limitations of Transformer architectures. RegFormer encodes each gene through dual embeddings, a value embedding for quantitative expression and a token embedding for regulatory identity, organized within a GRN-guided gene order to capture both expression dynamics and hierarchical regulation. Pretrained on 26 million human single cells spanning 45 tissues and diverse biological contexts, RegFormer achieves superior scalability and biological fidelity. Across comprehensive benchmarks, it consistently outperforms state-of-the-art single-cell foundation models (scGPT, Geneformer, scFoundation, and scBERT), delivering higher clustering accuracy, improved batch integration, and more precise cell type annotation. RegFormer also reconstructs biologically coherent GRNs, accurately models transcriptional responses to genetic perturbations, and enhances drug response prediction across cancer cell lines. By combining regulatory priors with efficient long-sequence Mamba modeling, RegFormer establishes a biologically grounded and scalable framework for single-cell representation learning, enabling deeper mechanistic insight into gene regulation and cellular state transitions.

bioinformatics↗

bioRxiv2025

Single-cell spatial transcriptome reveals pathological features of human hippocampus with sclerosis

More than half of epileptic patients ultimately turned to intractable epilepsy. Mesial temporal lobe epilepsy with hippocampal sclerosis (MTLE-HS), the most common type of intractable epilepsies, whose pathological mechanism remains elusive. Here, using 42 human hippocampal samples from surgical donors of MTLE (32 with and 10 without HS) through single-cell resolution Stereo-seq and histologcial experiments, we revealed spatially pathological changes of gene expression and cell type composition with HS systematically. After precise parcellation of hippocampal subregions and differentially expressed gene (DEG) analysis between each region with and without sclerosis, we found Cornu Ammonis (CA) subregions with higher number of DEGs were vulnerable to sclerosis, especially CA1 and CA3. Within CA1, we found that CA1-superficial and proximal areas were more vulnerable than CA1-deep and distal areas in sclerosis. Meanwhile, after analyzing 350,795 segmented cells from Stereo-seq, we found dramatically increasing density of astrocyte accompanied with significantly decreasing density of excitatory neurons in CA1, especially superficial and proximal CA1, and CA3 in sclerosis. In these two vulnerable subregions, proliferative astrocyte (P_astrocyte) and reactive astrocyte (R_astrocyte) were found to be enriched whereas apoptotic subtype of astrocyte (A_astrocyte), related with apoptotic pathway, was mainly located in alveus, which strengthened cell communication with reactive microglia (R_microglia) in HS, revealing the novel pathological feature in our work. The pseudotime analysis indicated that CA excitatory neurons underwent synaptic impairment, energy dysfunction, aging, and finally losing cell identity until death through autophagy or apoptosis. Besides, we also found a resilient subtype, EX_CA2-4.3, highly expressed extracellular matrix related genes including PDYN, and was increasing the interaction of BDNF-NTRK, NFASC-CNTN1 to withstand the damage from sclerosis. Together, our study provided a reference of human hippocampus with and without HS caused by MTLE, and highlighted the potential pathological mechanism on molecular and cellular level of MTLE-HS.

neuroscience↗

bioRxiv2025

Single cell resolved spatial immune repertoire unveils spatial heterogeneity of lymphoid aggregates in human immune disorders

Adaptive immunity, mediated by T and B cell responses, is essential for defending against infections and cancers while also being implicated in autoimmune diseases. Tracking T and B cell repertoires in situ at single-cell resolution is essential for understanding adaptive immune responses. To address the lack of tools for in situ single-cell T/BCR (XCR) sequencing, we developed Stereo-XCR-seq, an efficient strategy for retrieving and sequencing TCR and BCR from Stereo-seq cDNA libraries at subcellular resolution. Stereo-XCR-seq provides unbiased full-length XCR reads alongside spatial transcriptomics, enabling the identification of heterogeneous lymphoid aggregates with distinct clonal activities in cancers and inflammatory bowel disease (IBD). We identified plasma cell aggregates that differ from tertiary lymphoid structures (TLSs) in both transcriptomic profiles and clonal activities, with spatial positioning potentially mediating unique immune responses. Collectively, Stereo-XCR-seq enables in situ single-cell profiling of T and B cell clonal activities within tissue microenvironments, providing insights into lymphocyte adaption to environmental stimuli. This technology provides potential for advancing our understanding of tissue immunity and the development of therapeutic strategies for immune disorders.

immunology↗

bioRxiv2025

Genome-wide Viral Nascent RNA Sequencing Unveils Polymerase Pausing Landscape at Single-nucleotide Precision

Understanding viral replication and transcription mechanisms is critical for developing effective antiviral strategies. The study of viral gene regulation in host cellular environment is an important bridge for translating mechanistic discoveries from in vitro studies to in vivo but it remains stagnated due to the absence of technological advancement. Current methods for studying viral transcription and replication have been limited to capturing only the mature viral RNAs, obscuring the dynamic intermediates and mechanistic details of these crucial processes. Here, we introduce an original technology we called "Total Elongating Nascent VIral Polymerase single-molecule Sequencing (TenVIP-seq)", which isolates and analyses newly synthesized RNA within the viral RNA-dependent RNA polymerase (RdRp) complex, enabling the discovery of mechanisms critical for both viral replication and transcription. Our first characterization of nascent RNA species for an RNA virus showed that RdRP exhibited non-random pausing with profile signatures along the eight influenza A virus (IAV) gene segments. We also revealed genome-wide pausing at known regulatory sites, such as the poly(U) tract polyadenylation site, and new putative regulatory sites at single-nucleotide resolution. Distinct pausing features between the viral genomic (vRNA) and anti-genomic (cRNA/mRNA) templates were observed, suggesting that RdRp processes transcription and replication differentially on positive and negative sense RNA. The NTP analog drug T-705 (favipiravir) intensified RdRp pauses during genome replication and transcription without introducing new pausing sites, while TRIM25 knockout in host cells infected with virus reduced RdRp pausing globally across the viral genome. Strikingly, we observed that terminal nucleotide misincorporations of nascent RNA sequencing of paused RdRP, which were previously undetectable, were as high as an average of 51.6% (vRNA) and 44.0% (cRNA/mRNA), suggesting that the mutational rate of viruses is much higher than previously thought. We demonstrated that TenVIP-seq holds potential in providing insights into the molecular mechanisms that control viral genome replication, gene expression and regulation, mutation, antiviral drug treatment, and virus-host interaction.

genetics↗

bioRxiv2025

A comprehensive benchmark and guide for sequence-function interpretable deep learning models in genomics

The development of sequence-based deep learning methods has greatly increased our understanding of how sequence determines function. In parallel, numerous interpretable algorithms have been developed to address complex tasks, such as elucidating sequence regulatory syntax and analyzing non-coding variants from trained models. However, few studies have systematically compared and evaluated the performance and interpretability of these algorithms. Here, we introduce a comprehensive benchmark framework for evaluating sequence-to-function models. We systematically evaluated multiple models and DNA language foundation models using 369 ATAC-seq datasets, employing diverse training strategies and evaluation metrics to uncover their critical strengths and limitations. Our benchmark study highlights that different model architectures and interpretability methods are better suited to specific scenarios. Negative samples derived from naturally inactive regions outperform synthetic sequences, whereas single-cell tasks require specialized models. Additionally, we demonstrate that interpretable sequence-function models can complement traditional sequence alignment methods in studying cross-species enhancer regulatory logic. We also provide a pipeline to help researchers select the optimal sequence-function prediction and interpretability algorithms.

bioinformatics↗