Biology subjects

Jia, K.

Publications and source records attributed to Jia, K..

At least 19 recordsLinked to original sources

bioRxiv2026

From sensation to action: distributed mesoscale dynamics in the primate brain

Perceptual decisions engage distributed brain networks, yet how decision-related signals are organized across and within brain areas remains unclear. Mesoscale functional domains are established in visual cortex, but whether spatially distinct domains organize perceptual decisions across the brain is unknown. Here we used 7-T functional MRI at 1-mm isotropic resolution in two awake macaques performing a contrast-discrimination task to map brain-wide activity during sensation, decision formation and saccadic report. Voxel timecourses separated into three groups with distinct temporal profiles, associated with sensory, decision-related and action-related activity. Their spatial distributions formed a broad progression from visual cortex to parietal, prefrontal and motor regions, while revealing distinct response domains within these regions. Choice-related signals were distributed across sensory, associative, and motor systems before movement, including in visual cortex responses to physically identical stimuli that differed according to the animal's eventual choice. Moreover, within the intraparietal sulcus and other regions, mesoscale domains preferentially responsive to near-threshold conditions carried choice information under uncertainty, whereas domains responsive to high-contrast conditions carried choice information when evidence was strong. These findings reveal that perceptual decisions are organized across the brain through spatially differentiated mesoscale networks whose activity varies with task phase and sensory certainty. This spatial organization provides a framework for investigating how distributed neural signals support flexible choice.

neuroscience↗

bioRxiv2026

Combinatorial base editing couples disease correction with lineage amplification in hematopoietic stem and progenitor cells

First-generation genome editing therapies have largely focused on correcting or compensating for pathogenic variants. However, as these approaches enter the clinic, emerging biological constraints limit maximal therapeutic impact. Because globin genes are activated late during erythroid differentiation, genome-corrected hematopoietic stem and progenitor cells (HSPCs) gain little selective advantage in the bone marrow. Here, we establish a strategy that links therapeutic genome edits to an erythroid fitness-enhancing allele to amplify the output of clinically relevant cells. We develop a multiplex base editing strategy that couples fetal hemoglobin (HbF) reactivation with erythroid lineage expansion. Introduction of a naturally occurring erythropoietin receptor truncation (tEPOR) associated with benign erythrocytosis increased erythroid cell production without impairing viability or differentiation. Combinatorial editing of tEPOR together with the BCL11A erythroid enhancer and HBG1/2 promoters in healthy donor, sickle cell disease, and {beta}-thalassemia HSPCs synergistically increased erythroid proliferation and HbF expression beyond single base-edited or Casgevy-treated controls. Multiplex base-edited HSPCs retained long-term lineage repopulation and engraftment capacity in vivo, establishing a modular strategy that pairs disease correction with lineage amplification to improve therapeutic potency.

cell biology↗

bioRxiv2026

An ancient monoaminergic signaling system coordinates contractility in a nerveless sponge

Chemical neurotransmission was a key animal innovation, enabling multicellular coordination of physiology and locomotion. Sponges are early-diverging animals that lack neurons and muscles, yet still coordinate contraction and relaxation of their filter-feeding water canals. Here, we show that Spongilla lacustris synthesizes the monoamines tryptamine, phenethylamine, and tyramine to elicit distinct canal behaviors. We identify previously uncharacterized decarboxylases and vesicular transporters coexpressed in secretory neuroid and metabolic cells. Using phosphoproteomics and label-free 3D imaging, we show that tryptamine activates GPCR signaling and Rho GTPases, remodeling adhesion and actomyosin networks in contractile canal epithelia to drive localized constrictions and whole-body deflations. Together, these findings define an ancestral monoaminergic system linking secretory and contractile cell types that predate neurons and was later elaborated for neuromodulation of synaptic transmission.

evolutionary biology↗

bioRxiv2025

Reactivation-coupled brain stimulation enables complete learning generalization

Generalization of learned knowledge to new contexts is essential for adaptive behavior. Despite extensive research on the brain plasticity mechanisms underlying learning specificity, the mechanisms that facilitate generalization remain poorly understood. Here, we investigate whether using brain stimulation to disrupt offline consolidation in visual cortex promotes learning generalization. Separate groups of participants (N = 144) were trained on visual detection tasks using either a reactivation-based protocol or conventional full-practice, combined with anodal or sham transcranial direct current stimulation (tDCS) over the visual cortex. Strikingly, only combination of reactivation-based learning with anodal tDCS produced complete generalization from trained to untrained stimuli, an effect consistently replicated across features (orientation, motion direction). In contrast, reactivation-based learning alone and conventional full-practice - whether with or without brain stimulation - yielded stimulus-specific learning. Importantly, reactivation-coupled brain stimulation achieved generalization with an 80% reduction in training trials while maintaining learning gains comparable to full-practice. These findings demonstrate that reactivation and neuromodulation interact to unlock learning generalization, revealing a key brain plasticity mechanism and offering a rapid, translatable strategy for sensory rehabilitation.

neuroscience↗

bioRxiv2025

Methionine γ-lyase Traps Cofactor Pyridoxal-5'-phosphate through Specific Serine-Mediated Affinity

Pyridoxal-5'-phosphate (PLP), the ubiquitous and ancient cofactor, plays important roles in enzymatic elimination, transamination and other reactions. The catalytic efficiency of PLP-dependent enzymes is significantly higher than that of free PLP. The recruitment of PLP from the environment by the enzymes, particularly through interactions outside the active site, is the key step determining the occurrence of PLP-mediated catalysis. However, the precise mechanism by which enzymes recruit PLP remains elusive. Methionine {gamma}-lyase (MGL), a PLP-dependent enzyme, catalyzes the degradation of L-methionine, thereby suppressing cancer cell proliferation through serum or dietary methionine depletion. Here, we report the crystal structure of yMGL, which belongs to a newly identified subgroup of cystathionine {gamma}-lyases, in complex with L-methionine and PLP. Through truncating the C-terminal domain of yMGL both in vitro and in vivo, we demonstrated that this domain, outside the canonical PLP-binding domain, is essential for the specific interaction between yMGL and PLP, as well as for efficient L-methionine catabolism. A conserved Ser in the C-terminal domain, positioned at the active-site entrance of MGLs, confers specificity for PLP. These findings elucidate a previously uncharacterized mechanism of PLP recruitment by MGLs and promote rational design of MGLs with tailored cofactor selectivity and catalytic performance.

biochemistry↗

bioRxiv2025

Nervous Necrosis Virus Capsid Protein Functions as a SUMO E3 Ligase to Activate MAVS-Dependent NF-κB Signaling

Nervous necrosis virus (NNV) is a lethal aquatic pathogen that activates NF-{kappa}B signaling to manipulate host immune responses, yet the underlying molecular mechanisms remain poorly defined. Here, we identify the NNV capsid protein (CP) as a novel viral SUMO E3 ligase that promotes SUMOylation of Lateolabrax japonicus mitochondrial antiviral signaling protein (MAVS) to drive NF-{kappa}B activation. We show that CP induces p65 nuclear translocation and upregulates proinflammatory cytokine expression in a MAVS-dependent manner. Mechanistically, CP interacts with SUMO2 and the E2 conjugating enzyme L. japonicus UBC9 to promote MAVS SUMOylation at lysine 325 (K325), which stabilizes MAVS, facilitates its aggregation, and amplifies downstream NF-{kappa}B signaling. Notably, disruption of this modification via the K325R mutation or SUMOylation inhibition abrogates MAVS aggregation and inflammatory responses. Our findings reveal a previously unrecognized strategy by which NNV hijacks the host SUMOylation machinery to fine-tune MAVS function, promoting immune evasion and viral persistence. SummaryNervous necrosis virus (NNV) is a highly pathogenic virus in aquatic animals that manipulates host immune signaling to promote infection. Here, we identify the NNV capsid protein (CP) as a viral SUMO E3 ligase that modifies the mitochondrial antiviral signaling protein MAVS via SUMOylation at lysine 325. This modification enhances MAVS stability and aggregation, leading to robust activation of the NF-{kappa}B signaling pathway and increase expression of proinflammatory cytokines. Disruption of this modification abolishes CP-mediated immune activation. This study unveils a previously unrecognized mechanism by which NNV exploits host SUMOylation machinery to modulate MAVS function, contributing to immune evasion and facilitating viral persistence. These findings provide valuable insights into virus-host interactions and highlight potential therapeutic targets against NNV infections.

microbiology↗

bioRxiv2025

UNCOVERseq Enables Sensitive and Controlled Gene Editing Off-Target Nomination Across CRISPR-Cas Modalities and Systems

The rapid development of CRISPR-Cas gene editing technologies has revolutionized genetic medicine, offering unprecedented precision and potential for treating a wide array of genetic disorders. However, assessing the risks of unintended gene editing effects remains critical, and is complicated by new editing modalities and unclear analytical guidelines. We present UNCOVERseq (Unbiased Nomination of CRISPR Off-target Variants using Enhanced RhPCR), an improved in cellulo off-target nomination workflow designed to sensitively nominate off-target sites (<0.01% editing) with defined input requirements and analytical process controls to provide empirical performance evidence across diverse circumstances. Using this workflow, we nominated off-targets across 192 guide RNAs (gRNAs) and demonstrated superior performance compared to existing methodologies. We identified a subset of six gRNAs with a dynamic range of specificity and confirmed the relevance and high true positive rate of our nomination method, providing relative risk assessments for multiple modalities (S.p. Cas9 and derived high-fidelity variants / base editors) in a translational system involving hematopoietic stem and progenitor cells (HSPCs). Additionally, we established that double-strand break (DSB) editing retains a strong, positive rank correlation to single-strand break (SSB)-mediated base editing, highlighting the importance of DSB nomination sites as candidate loci for base editing. Overall, UNCOVERseq improves informed risk assessment of gene editing in translational systems by enhancing the quality of off-target nomination.

molecular biology↗

bioRxiv2025

Conserved and lineage-restricted gene regulatory programs modulate developmental cnidocyte specification in Nematostella vectensis.

Cnidocytes are a synapomorphy of cnidarians that have evolved a range of morphologies and functions within and across extant species, which makes them an excellent model to investigate how novel cell types emerge and radiate in evolution. One way to gain insight into how cell types evolve is to investigate the gene regulatory networks (GRNs) that pattern them, leading to the identification of shared and divergent programs that likely represent components of the ancestral and evolved phenotypes. Efforts to identify early acting transcription factors in the sea anemone Nematostella vectensis revealed that NvfoxE-like is a lineage-restricted cnidocyte transcription factor. Expression and functional studies confirmed that NvfoxE-like and its targets are found exclusively in cnidocytes. By investigating interactions with regulators of the known cnidogenesis GRN and comparing the targets of known cnidocyte regulators, we identified at least four regulatory programs contributing to the cnidogenic GRN in Nematostella. The previously identified cnidocyte regulators Nvznf845 and NvpaxA represent two core pan-cnidarian regulatory programs, while the NvfoxE-like dependent program appears to be restricted to Hexacorallia or Actinaria, where we hypothesize it functions to pattern the unique capsule features present in those lineages. Here, we confirm the hypothesized core conserved transcriptional program, as well as demonstrate how a novel transcription factor was co-opted into cnidogenesis in a subset of cnidarians, thus contributing to the diversification and continued evolution of cnidocytes.

evolutionary biology↗

bioRxiv2024

Myosin light chain 3 serves as a receptor for nervous necrosis virus entry into host cells via the macropinocytosis pathway

Nodaviridae infections cause severe mortality in insects and fish, with nervous necrosis virus (NNV) posing significant threats to global fish populations. However, the host factors involved in NNV entry remain poorly understood. We identify myosin light chain 3 from marine medaka (Oryzias melastigma) (MmMYL3) as a novel receptor for red-spotted grouper NNV (RGNNV), facilitating internalization via macropinocytosis. MmMYL3 directly binds the RGNNV capsid protein (CP), which dependents on the arm and S domains of CP and the EF-hand2 domain of MmMYL3. In vitro experiments showed that MmMYL3 siRNA, protein, anti-MYL3 antibodies or the arm domain synthetic peptides blocked RGNNV entry. Moreover, in vivo administration of MmMYL3 protein also inhibited RGNNV infection. Ectopic MmMYL3 expression enabled RGNNV internalization into resistant cells. Notably, MmMYL3 facilitated RGNNV internalization through the macropinocytosis pathway via the IGF1R-Rac1/Cdc42 axis. Collectively, our findings underscore MYL3s crucial role in NNV entry and its potential as an antiviral target.

microbiology↗

bioRxiv2024

Dual-format attentional template during preparation in human visual cortex

Goal-directed attention relies on forming internal templates of key information relevant for guiding behavior, particularly when preparing for upcoming sensory inputs. However, evidence on how these attentional templates is represented during preparation remains controversial. Here, we combine functional magnetic resonance imaging (fMRI) with an orientation cueing task to isolate preparatory activity from stimulus-evoked responses. Using multivariate pattern analysis, we found decodable information of the to-be-attended orientation during preparation; yet preparatory activity patterns were different from those evoked when actual orientations were perceived. When perturbing the neural activity by means of a visual impulse (pinging technique), the preparatory activity patterns in visual cortex resembled those associated with perceiving these orientations. The observed differential patterns with and without the impulse perturbation suggest a predominantly non-sensory format and a latent, sensory-like format of representation during preparation. Furthermore, the emergence of the sensory-like template coincided with enhanced information connectivity between V1 and frontoparietal areas and was associated with improved behavioral performance. By engaging this dual-format mechanism during preparation, the brain is able to encode both abstract, non-sensory information and more detailed, sensory information, potentially providing advantages for adaptive attentional control. For example, consistent with recent theories of visual search, a predominantly non-sensory template can support the initial guidance and a latent sensory-like format can support prospective stimulus processing.

neuroscience↗

bioRxiv2024

Phytop: A tool for visualizing and recognizing signals of incomplete lineage sorting and hybridization using species trees output from ASTRAL

Incomplete lineage sorting (ILS) and introgression/hybridization (IH) is prevalent in nature and thus frequently result in discrepancies within phylogenetic tree topologies, leading to misinterpretation of phylogenomic data. Despite the availability of numerous tools for detecting ILS and IH among species, many of these tools are lacking effective visualization, or are time-consuming, or require prior predetermination. Here, we addressed these shortcomings by developing a fast-running, user-friendly tool called Phytop. By defining ILS and IH indices to quantify ILS and IH, this tool can detect the extent of ILS and IH among lineages with high reliability, and can visualize them based on the gene tree topology patterns constructed using ASTRAL. We tested Phytop extensively using both simulated and real data, and found that it enables users to quickly and conveniently estimate the extent of ILS and IH, thus clarifying the phylogenetic uncertainty. Phytop is available at https://github.com/zhangrengang/phytop and is expected to conveniently contribute to the intuitive inference of genetic relationships among lineages in future research.

bioinformatics↗

bioRxiv2023

Noise reduction as a unified mechanism of perceptual learning in humans, macaques, and convolutional neural networks

Visual perceptual learning (VPL), defined as long-term improvement in a visual task, is considered a crucial tool for elucidating underlying visual and brain plasticity. However, the identification of a unified theory of VPL has long been controversial. Multiple existing models have proposed diverse mechanisms, including improved signal-to-noise ratio, changes in tuning curves, and reduction of noise correlations, as major contributors to improved neural representations associated with VPL. However, each model only accounts for specific aspects of the empirical findings, and there exists no theory that can comprehensively explain all empirical results. Here, we argue that all neural changes at single units can be conceptualized as geometric transformations of population response manifolds in a high-dimensional neural space. This approach enables conflicting major models of VPL to be quantitatively tested and compared within a unified computational theory. Following this approach, we found that changes in tuning curves and noise correlations, as emphasized by previous models, make no significant contributions to improved population representations by visual training. Instead, we identified neural manifold shrinkage due to reduced trial-by-trial neural response variability, a previously unexplored factor, as the primary mechanism underlying improved population representations. Furthermore, we showed that manifold shrinkage successfully accounts for learning effects across various domains, including artificial neural responses in deep neural networks trained on typical VPL tasks, multivariate BOLD signals in humans, and multi-unit activities in monkeys. These converging results suggest that our neural geometry theory offers a quantitative and comprehensive approach to explain a wide range of empirical results and to reconcile previously conflicting models of VPL.

neuroscience↗

bioRxiv2023

Dual α-globin and truncated erythropoietin receptor knock-in restores hemoglobin production in α-thalassemia major-derived hematopoietic stem and progenitor cells

Alpha-thalassemia is an autosomal recessive disease with increasing worldwide prevalence. The molecular basis is due to mutation or deletion of one or more duplicated -globin genes, and disease severity is directly related to the number of allelic copies compromised. The most severe form, -thalassemia major (TM), results from loss of all four copies of -globin and has historically resulted in fatality in utero. However, in utero transfusions now enable survival to birth. Postnatally, patients face challenges similar to {beta}-thalassemia, including severe anemia and erythrotoxicity due to imbalance of {beta}-globin and -globin chains. While curative, hematopoietic stem cell transplantation (HSCT) is limited by donor availability and potential transplant-related complications. Despite progress in genome editing treatments for {beta}-thalassemia, there is no analogous curative option for patients suffering from -thalassemia. To address this, we designed a novel Cas9/AAV6-mediated genome editing strategy that integrates a functional -globin gene into the {beta}-globin locus in TM patient-derived hematopoietic stem and progenitor cells (HSPCs). Incorporation of a truncated erythropoietin receptor transgene into the -globin integration cassette dramatically increased erythropoietic output from edited HSPCs and led to the most robust production of -globin, and consequently normal hemoglobin. By directing edited HSPCs toward increased production of clinically relevant RBCs instead of other divergent cell types, this approach has the potential to mitigate the limitations of traditional HSCT for the hemoglobinopathies, including low genome editing and low engraftment rates. These findings support development of a definitive ex vivo autologous genome editing strategy that may be curative for -thalassemia. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=177 SRC="FIGDIR/small/555926v2_ufig1.gif" ALT="Figure 1"> View larger version (41K): org.highwire.dtl.DTLVardef@1b63dacorg.highwire.dtl.DTLVardef@18b25a1org.highwire.dtl.DTLVardef@53835corg.highwire.dtl.DTLVardef@d52bf5_HPS_FORMAT_FIGEXP M_FIG C_FIG

molecular biology↗

bioRxiv2023

Viruses in deep-sea cold seep sediments harbor diverse survival mechanisms and remain genetically conserved within species

Deep sea cold seep sediments have been discovered to harbor novel, abundant, and diverse bacterial and archaeal viruses. However, little is known about viral genetic features and evolutionary patterns in these environments. Here, we examined the evolutionary ecology of viruses across active and extinct seep stages in the area of Haima cold seeps in the South China Sea. Diverse antiviral defense systems in 487 microbial genomes spanning 43 families inform the dynamics of host-virus interactions. Accordingly, a total of 338 viral operational taxonomic units are identified and linked to 36 bacterial and archaeal phyla. Cold seep viruses are predicted to harbor diverse adaptive strategies to persist in this environment, including counter-defense systems, reverse transcriptases, auxiliary metabolic genes, and alternative genetic code assignments. Extremely low nucleotide diversity is observed in cold seep viral populations, being influenced by factors including microbial host, sediment depth and cold seep stage. Most cold seep viral genes are under strong purifying selection with trajectories that differ depending on whether cold seeps are active or extinct. This work sheds light on the understanding of environmental adaptation mechanisms and evolutionary patterns of viruses in the sub-seafloor biosphere.

microbiology↗

bioRxiv2022

Ring-finger protein 34 facilitates nervous necrosis virus evading antiviral innate immunity by targeting TBK1 and IRF3 for ubiquitination and degradation

Ubiquitination, as one of the most prevalent posttranslational modifications of proteins, enables a tight control on host immune responses. Many viruses hijack the host ubiquitin system to regulate host antiviral responses for their survival. Here, we found that fish pathogen nervous necrosis virus (NNV) recruited an E3 ubiquitin ligase ring finger protein 34 (RNF34) to inhibit RLRs-mediated interferons (IFN) response via ubiquitinating TBK1 and IRF3. Ectopic expression of RNF34 greatly enhances NNV replication and prevents IFN production, while deficiency of RNF34 led to the opposite effect. Furthermore, RNF34 targets TBK1 and IRF3 via its RING domain. Of note, the interactions between RNF34 and TBK1 or IRF3 were conserved in different fish species. Mechanically, RNF34 promote K27-linked ubiquitination and degradation of TBK1 and IRF3, which in turn diminishing TBK1-induced translocation of IRF3 from cytoplasm to nucleus. Ultimately, NNV capsid protein (CP) was found directly bind with RNF34 and this interaction was conserved in different fishes, and CP induced TBK1 and IRF3 degradation and IFN suppression was depended on RNF34. Our finding demonstrated a novel mechanism by which NNV CP evaded host innate immunity via RNF34, and provided a potential drug target for the control of NNV infection. Author SummaryUbiquitination plays an essential role in the regulation of innate immune responses to pathogens. NNV, a kind of RNA virus, is the causal agent of a highly destructive disease in a variety of marine and freshwater fish. Previous study reported NNV could hijack the ubiquitin system to manipulate the hosts immune responses, however, how NNV utilizes ubiquitination to facilitate its own replication is not well understood. Here, we identified a novel distinct role of E3 ubiquitin ligase RNF34 as an IFN antagonist to promote NNV infection. Nervous necrosis virus capsid protein utilized RNF34 to target TBK1 and IRF3 for K27 and K48-linked ubiquitination degradation. Importantly, the interactions between RNF34 and CP, TBK1 or IRF3 are conserved in different fishes, suggesting it is a general immune evasion strategy exploited by NNV to target the IFN response via RNF34.

immunology↗

bioRxiv2022

A novel inovirus reprograms metabolism and motility of marine Alteromonas

Members from the Inoviridae family with striking features are widespread, highly diverse and ecologically pervasive across multiple hosts and environments; however, very small amount of inoviruses have been isolated and studied. Here, a filamentous phage infecting Alteromonas abrolhosensis, designated {phi}AFP1, was isolated from the South China Sea and represented as a novel genus of Inoviridae. {phi}AFP1 consisted of a single-stranded DNA genome (5986 bp), encoding eight putative ORFs. Comparative analyses revealed {phi}AFP1 could be regarded as genetic mosaics, which especially came from Ralstonia and Stenotrophomonas phages. The temporal transcriptome analysis of A. abrolhosensis to {phi}AFP1 infection reveals that 7.78% of the host genes were differentially expressed. The genes involved in translation processes, ribosome pathways and degradation of multiple amino acid pathways at plateau period were upregulated, while host material catabolic and bacterial motility-related genes were downregulated, indicating that {phi}AFP1 might hijack the energy of the host for the synthesis of phage proteins. {phi}AFP1 exerted the step-by-step control on host genes through the appropriate level of the utilizing host resources, affirming a new non-standard regulatory strategy of viral temperately control over the host transcriptional profile. Our study provides novel information for a better understanding of filamentous phage characteristics and phage-host interactions.

microbiology↗

bioRxiv2022

Evolutionary ecology of microbial populations inhabiting deep sea sediments associated with cold seeps

Deep sea cold seep sediments host abundant and diverse bacterial and archaeal populations that significantly influence biogeochemical cycles. While numerous studies have revealed the community structure and functional capabilities of cold seep microbiomes, little is known about their genetic heterogeneity within species. Here, we examined intraspecies diversity patterns of 39 abundant species identified in sediment layers down to 4.3 mbsf across six cold seep sites from around the world. These species were predicted to participate in methane oxidation and sulfate reduction, and based on their metabolic capabilities, grouped as aerobic methane-oxidizing bacteria (MOB), anaerobic methanotrophic archaea (ANME) and sulfate-reducing bacteria (SRB). These physiologically and phylogenetically diverse MOB, ANME and SRB display different degrees of intrapopulation sequence divergence and different evolutionary trajectories. Populations were in general characterized by low rates of homologous recombination and strong purifying selection with most of the nucleotide variation being synonymous. Functional genes related to methane (pmoA and mcrA) and sulfate (dsrA) metabolisms were found to be under strong purifying selection in the vast majority of species investigated, although examples of active positive selection were also observed. These genes differed in evolutionary trajectories across phylogenetic clades but are functionally conserved across cold seep sites. Intrapopulation diversification of MOB, ANME and SRB species as well as their mcrA and dsrA genes was observed to be depth-dependent and undergo divergent selection pressures throughout the sediment column. These results highlight the role of the interplay between ecological processes and the evolution of key bacteria and archaea in deep sea cold seep sediments and shed light on how microbial populations adapt in the subseafloor biosphere.

microbiology↗

bioRxiv2022

Protein Language Model Performs Efficient Homology Detection

MotivationThere are now 225 million sequences in the UniProtKB database, as of January 2022 and 451 million protein sequences in the NCBI non-redundant database. This huge sequence data is ripe for analysis and can be extremely informative about biological function when analyzed with the appropriate methods. Evolutionary information such as the relationship among protein sequences is key to performing sequence analyses. Since sequence matching is one of the primary ways that annotations are found, higher-quality sequence matches yield a larger number of identified homologs. Thus, there is an essential need for a faster and more accurate homolog detection method to process the huge amount of rapidly growing biological sequences. MethodRecently, we have seen major improvements in various predictive computational tasks such as structure prediction from the ever-improving artificial intelligence methods. One such approach has been to use language models to represent proteins numerically in a representation matrix (embeddings) while retaining context-dependent biochemical, biophysical, and evolutionary information. Computational transformer architectures that utilize attention neural networks can generate these context-aware numerical representations in an unsupervised fashion. One such use for these protein embeddings is remote homolog detection. In this work, we utilize protein language models and then apply discrete cosine transforms to extract the essential part of these embeddings, resulting in a significantly smaller fixed-size matrix for each sequence. This allows us to numerically and efficiently calculate the distance between all pairs of proteins resulting in homolog detection. ResultsOur Protein LAnguage model Search Tool (PLAST) is significantly faster, with linear runtimes in the number of sequences within the query database. With only one CPU core, it can scan a million sequences in less than a second. It essentially removes the noise in the sequence data and leads to significant improvements. PLAST is more accurate in the benchmarks tested from the PFAM, SCOP, and CATH databases than other approaches. When benchmarked with the PFAM database, the increase in the area under the receiver operating characteristic curve (AUROC) 3.1% when compared with NCBI-BLAST. The number of remote homologs that are detectable now is significantly larger and pushes sequence matches deeply into the usual twilight zone. Compared with the state-of-the-art profile-based homology search tools like CSBLAST, the increase was still 2.0%. PLAST can find remote homologs for a significant number of proteins that had been thought to be unique due to homolog detection failure. These homologs that are found usually have less than 20% sequence identity making them indistinguishable from noise with most other sequence matching methods. ConclusionPLAST is an accurate and fast homolog detection tool essential for easy and rapid progress to utilize the vast amount of data generated by next-generation sequencing methods. Quantization of sequence embeddings into highly-compressed noise-free representations with the use of direct cosine transforms allows for the efficient and accurate detection of normal homologs and remote ones that are un-detectable by other sequence similarity methods. The PLAST web server is accessible from https://mesihk.github.io/plast.

bioinformatics↗