bioRxiv Science⌕ Search

Biology subjects

CHEN, L.

Publications and source records attributed to CHEN, L..

10 recordsLinked to original sources
bioRxiv2026

Client-server interfaces enable efficient agent-driven variant calling

BackgroundLarge language model (LLM) agents increasingly automate bioinformatics analyses, but most existing bioinformatics tools were built for standalone use by human experts. An agent driving such a tool must reason about its installation, configuration, and execution from documentation for human, spending many turns, tokens, and tool calls per result. How a method is exposed to an agent can therefore matter as much as the method itself. By designing agentic interfaces for these tools, agent can reduce such overhead and improve the reliability of agent-driven analyses. FindingsTo test this design, we re-architected Clair3, a widely used deep-learning-based long-read variant caller, into a client-server system, Clair3-Connect. The client performs all genomics related processing and holds the identifiable data. The server runs only neural-network inference, and the client sends only feature tensors to the server, while sample identifiers and genomic context remain on the client. The client exposes schema-defined agent-facing tools that an agent invokes through single structured calls. On an APOE diplotyping task, all 60 agent runs were correct. The agentic tools used 12K tokens in 3 turns, 6.8 to 14 times fewer tokens than the shell-driven baselines (81K-163K tokens), at about a quarter the wall-clock time and far more stably (4% versus 35% token usage variation). Dropping the pileup and phasing stages to keep the client light left SNP F1 within 0.1-0.3 points of standard Clair3 by 50x coverage, while mutual TLS and AES-256-GCM encryption added 7.2% to end-to-end runtime. ConclusionsRecasting an established algorithm as developer-built, agentic tools behind a secure client-server boundary makes it more efficient, reliable, and easier to deploy for an LLM agent than a third-party wrapper, which cannot recover the defaults and conventions only its developers know. Agentic interfaces should be a first-class deliverable of bioinformatics tool development.

bioinformatics↗
bioRxiv2025

Identification and characterization of a novel exo-glycosidase activity of NGLY1 on ENGase-digested N-GlcNAc proteins in vitro

N-Glycanase 1 (NGLY1) is known for its activity to completely remove conjugated N-glycans from glycoproteins. Early studies illustrated that NGLY1 defects were associated with a rare disease named NGLY1-related congenital disorder of deglycosylation (NGLY1-CDDG). Although extensive research has been conducted over the past decade on the biological impact of NGLY1s endo-(N-)glycanase activity in cells and on disease, whether NGLY1 also exhibits an exo-(N-)glycanase activity remains open. In this study, an exo-(N-)glycanase activity of NGLY1 on the N-GlcNAc proteins was firstly reported and characterized in vitro. To distinguish for the conventional endo-(N-)glycanase activity and the newly reported exo-(N-)glycanase activity of NGLY1, the active sites of exo-NGLY1 were predicted and compared with those of endo-NGLY1. In addition, the correlation between NGLY1s exo-(N-)glycanase activity in vitro and NGLY1-CDDGs clinical characterization was investigated and discussed. The observation of NGLY1s novel exo-(N-)glycanase activity expanded the functional repertoire of NGLY1 beyond canonical endo-(N-)glycanase-mediated deglycosylation, and may provide a new framework to explain the clinical heterogeneity of NGLY1-CDDG.

biochemistry↗
bioRxiv2025

Biologically Informative NA Deconvolution (BIND) excavates hidden features of the proteome from missing values in large-scale datasets

The fast-advancing mass spectrometry and related technologies have greatly extended the depth of coverage in large-scale proteomics studies, including single-cell applications. As sample numbers grow rapidly, it is often challenging to interpret the proteins with missing values that are often presented as "NA" (not available). It could be the evidence of no expression, low expression below the detection threshold, or false negative detection due to technical issues. Existing methods for missing values imputation, while generally useful, rarely consider the non-random NA values that inform biological significance. In the current study, we developed Biologically Informative NA Deconvolution (BIND) that applies an adaptive neighborhood-based modeling to deconvolve the nature of NAs as "biological" (low/no expression) or technical (experimental errors). Applying to multiple cell line datasets and human tissue extracellular vesicle datasets, BIND excavated the NAs that indicated "hallmark absence" of unique proteins. This led to improvements in protein-protein interaction analysis and the identification of novel disease biomarkers. To facilitate its public accessibility, we compiled BIND into a web server that features functional online operations and interactive visualizations. Furthermore, we demonstrated that the BIND server could deconvolve the NAs and improve the analyses of single-cell proteomics datasets. Overall, BIND delineates the biological significance of missing values rather than treating them as a burden, providing a critical perspective for understanding the complex proteome in various biological contexts. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=144 SRC="FIGDIR/small/660508v1_ufig1.gif" ALT="Figure 1"> View larger version (52K): org.highwire.dtl.DTLVardef@1a9d7c4org.highwire.dtl.DTLVardef@194b2d9org.highwire.dtl.DTLVardef@169d3bborg.highwire.dtl.DTLVardef@cba0de_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioinformatics↗
bioRxiv2025

Discovery and analysis of an 841 kbp phage genome: the largest known to date

Viruses represent the most abundant biological entities on Earth, with bacteriophages (or phages) specifically infecting bacteria. Typically, double-strand DNA phages possess genomes around 50 kilobase pairs (kbp) in size, although the largest known genome surpasses 735 kbp. This raises intriguing questions about the potential maximum size of phage genomes. In this study, we present the first phage genome over 800 kbp, named BF3, which was reconstructed from oil reservoir production water. BF3 encodes 1,164 protein-coding genes (code 11) and 46 tRNA genes, and is classified within the Caudoviricetes, though its bacterial host remains unidentified. We utilized ColabFold to predict the structures of 744 protein-coding genes with high confidence, finding that 395 and 591 of these genes corresponded with known structures in the Protein Data Bank and the Big Fantastic Virus Database (BFVD), respectively. Notably, 153 of BF3s predicted structures exhibited no similarity to those catalogued in the BFVD, which is currently the most extensive viral protein structure database. This study not only expands our understanding of phage genome capacities but also underscores the need for specialized analytical tools and pipelines to investigate exceptionally large phages.

microbiology↗
bioRxiv2024

A nanobody-enzyme fusion protein targeting PD-L1 and sialic acid exerts anti-tumor effects by affecting tumor associated macrophages

Cancer cells employ various mechanisms to evade immune surveillance. Their surface features, including a protective "sugar coat" and immune checkpoints like PD-L1 (programmed death ligand 1), can impede immune cell recognition. Sialic acids, which carry negative charges, may hinder cell contact through electrostatic repulsion, while PD-L1 transmits immunosuppressive signals to T cells. Furthermore, cancer cells manipulate macrophages within the tumor microenvironment to facilitate immune escape. Prior research has demonstrated the effectiveness of separately blocking the PD-L1 and sialic acid pathways in eliciting anti-tumor effects. In this study, we investigated the relationship between PD-L1 expression and genes associated with sialic acid in clinical databases. Subsequently, we developed a novel nanobody enzyme fusion protein termed Nb16-Sia to simultaneously target both PD-L1 and sialic acid pathways. In vivo experiments confirmed the anti-tumor activity of Nb16-Sia and highlighted its dependence on macrophages. Further investigations revealed that Nb16-Sia could polarize macrophages towards the M1 phenotype through the C-type lectin pathway in vitro and eliminate tumor-associated macrophages in vivo. In conclusion, our findings demonstrate that the fusion of PD-L1 nanobody with sialidase effectively targets tumor-associated macrophages, resulting in significant anti-tumor effects. This approach holds promise for drug development aimed at enhancing immune responses against cancer.

pharmacology and toxicology↗
bioRxiv2023

Identification and characterization of a beta-1,4 Galactosidase from Elizabethkingia meningoseptica and its application on living cell surface

The biological function of terminal galactose on glycoprotein is an open field of research. Although progress had being made on enzymes that can remove the terminal galactose on glycoproteins, there is a lack of report on galactosidases that can work directly on living cells. In this study, a unique beta 1,4 galactosidase was isolated from Elizabethkingia meningoseptica (Em). It exhibited favorable stability at various temperatures (4-37) and pH (5-8) levels and can remove {beta}-1, 4 linked galactoses directly from glycoproteins. Using Alanine scanning, we found that two acidic residues (Glu-468, and Glu-531) in the predicted active pocket are critical for galactosidase activity. In addition, we also demonstrated that it could cleave galactose residues present on living cell surface. As the enzyme has a potential application for living cell glycan editing, we named it glycan editing galactosidase I or geGalaseI. In summary, our findings lay the groundwork for prospective investigations by presenting a prompt and gentle approach for the removal of galactose moieties from cell surface.

biochemistry↗
bioRxiv2022

A full spectrum PNGase activity analysis of R328 mutations on NGLY1

In humans, N-glycanase 1 (NGLY1; Peptide: N-glycanase, PNGase) is responsible for the deglycosylation of misfolded glycoproteins. Pathogenic mutations in NGLY1 cause a clinical condition known as congenital disorder of deglycosylation (NGLY1-CDDG), a rare autosomal recessive disease first reported in 2012. Although NGLY1-CDDG was diagnosed through whole-exome or whole-genome sequencing and by evaluating the expression levels of NGLY1, the clinical relevance of a detected mutation in NGLY1 needs to be further confirmed. In this study, an in vitro enzymatic assay system was established to evaluate the thermal stability and substrate specificity of NGLY1, as well as the optimum reaction conditions for its activity. A panel of all mutations at the amino acid site R328 in NGLY1 was subjected to this assay. The results revealed that R328A, R328D, R328E, R328F, R328G, R328I, R328P, R328V, R328W, and R328Y were dysfunctional mutations (10/19); NGLY1 mutations with R328H and R328T exhibited similar activity as wild-type NGLY1 (2/19); and NGLY1 mutations with R328C, R328K, R328L, R328M, R328N, R328Q, and R328S showed decreased activity (7/19) compared to wild-type NGLY1. In addition, the effect of potential regulatory compounds, including N-acetyl-L-cysteine and dithiothreitol, on NGLY1 was examined. This in vitro assay may serve as a standard protocol to facilitate rapid diagnosis of all mutations on NGLY1 and a practical screening method for drugs and compounds with potential therapeutic value for NGLY1-CDDG patients.

biochemistry↗
bioRxiv2021

Single Cell Raman-Deuterium Isotope Probing for Drug Resistance of Elizabethkingia spp.

Nosocomial infection associated with Elizabethkingia spp. is an emerging clinical concern characterized by multi-drug resistance and severe clinical consequences particularly in immunocompromised individuals and infants. Efficient control of this infection demands quick and reliable methods to determine the right drugs for the treatment. In this study, E. meningoseptica ATCC 13253 and four clinical isolates of Elizabethkingia spp. obtained from China, were subjected to single cell Raman spectroscopy analysis coupling with deuterium probing (single cell Raman-DIP). The results demonstrated that single cell Raman-DIP could generate an antimicrobial susceptibility testing result for Elizabethkingia spp. colonies within 4 hours based on their metabolisms variations at single cell level, and the drug resistant spectra of Elizabethkingia spp. determined by single cell Raman-DIP were consistent with the classical MIC method. Meanwhile single cell Raman spectroscopy (single cell RS) was applied to analyze Raman spectra of Elizabethkingia spp., which were revealed that their ratios of nucleic acid/protein were lower than other gram-negative pathogens and isolates from different origins could be distinguished by their Raman fingerprint. The in vitro results confirmed that minocycline and levofloxacin are first-line antimicrobials for Elizabethkingia spp. infection.

microbiology↗
bioRxiv2019

Self-reported sleep duration and daytime napping are associated with renal hyperfiltration and microalbuminuria in apparently healthy Chinese population

BackgroundSleep duration affects health in various way. The objective of this study was to investigate the relationship between sleep duration, daytime napping and kidney function in a middle-aged apparently healthy Chinese population.\n\nMethodsAccording to self-reported total sleep and daytime napping duration, 33,850 participants aged 38 to 90 years old from 8 regional centers were divided into subgroups. Height, weight, waistline, hipline, blood pressure, biochemical index, FBG, PBG, HbA1c, creatinine and urinary albumin-creatinine ratio (UACR) were measured and recorded in each subject. Microalbuminuria was defined as UACR>=30 mg/g, CKD was defined as eGFR<60 ml/min and hyperfiltration was defined as eGFR>=135 ml/min. Multiple logistic regressions were applied to investigate associations between sleep and kidney function.\n\nResultsCompared to participants with [7-8]h/day sleep, ORs of >9 h/day, (8, 9]h/day and <6h/day sleep for microalbuminuria were 1.317 (1.200-1.446, p<0.001), 1.215 (1.123-1.315, p<0.001) and 1.218 (0.967-1.534, p=0.094). eGFR levels were U-shaped associated with sleep duration among subjects with >=90ml/min eGFR, and N-shaped associated with sleep duration among subjects with <90ml/min eGFR. OR of >9h/day sleep for hyperfiltration was 1.400 (1.123-1.745, p=0.003) among eGFR>=90 ml/min participants. Daytime napping had a negative effect on renal health. Compared to participants did not have napping habit, the ORs of (0, 1]h/day, (1, 1.5]h/day and >1.5h/day daytime napping for microalbuminuria were 1.477 (1.370-1.591, p<0.001), 1.217 (1.056, 1.403, p=0.007) and 1.447 (1.242, 1.687, p<0.001).\n\nConclusionsTotal sleep duration are U-shaped associated with renal health outcomes. Daytime napping had a negative effect on renal health.

epidemiology↗