Biology subjects

Guo, A.-Y.

Publications and source records attributed to Guo, A.-Y..

6 recordsLinked to original sources

bioRxiv2026

OmniTCR: a foundation model unifying T cell receptor recognition prediction and conditional sequence generation

T cell receptor (TCR) recognition prediction and receptor generation are traditionally modelled separately, leaving vast TCR sequence collections disconnected from smaller TCR-peptide-MHC datasets. Here we present OmniTCR, a 113-million-parameter autoregressive foundation model pretrained on 328 million formatted human immune-sequence records. Sequence-type tokens and complementary component orders enable joint learning from individual TCR chains and partial or complete TCR-pMHC associations. On unseen epitopes, OmniTCR achieved AUPRCs of 0.7009 for peptide-TCR{beta}; recognition and 0.8235 for TCR-pMHC interaction prediction, exceeding the strongest evaluated comparators by 0.3396 and 0.3451, respectively. It distinguishes cancer from healthy repertoires across 11 independent pan-cancer cohorts (mean AUROC, 0.9436). The model achieved the highest sequence recovery on internal and external generation benchmarks. Structural modelling supported the plausibility of selected pMHC-conditioned CDR3{beta}; candidates. OmniTCR bridges heterogeneous immune sequence data, providing a foundation for computational immunology and receptor design.

bioinformatics↗

bioRxiv2023

Role of C2-domain protein CAR1 in the plasma membrane ABA signaling response

Plasma membrane (PM)-associated abscisic acid (ABA) signal transduction is an important component of ABA signaling. The C2-domain ABA-related (CAR) proteins have been reported to play a crucial role in recruiting ABA receptor PYR1/PYL/RCAR (PYLs) to the PM. However, the molecular details of the involvement of CAR proteins in membrane-delimited ABA signal transduction remain unclear. Here, the GUS-tagged materials for all Arabidopsis CAR members were used to comprehensively visualize the extensive expression patterns of the CAR family genes. Based on the specificity of CAR1s response to ABA, we determined to use it as a target to study the function of CAR proteins in PM-associated ABA signaling. Single-particle tracking showed that ABA affected the spatiotemporal dynamics of CAR1. The presence of ABA prolonged the dwell time of CAR1 on the membrane and showed faster lateral mobility. Surprisingly, we verified that CAR1 could directly recruit Hypersensitive to ABA1 (HAB1) and SNF1-related protein kinase 2.2 (SnRK2.2) to the PM at both the bulk and single-molecule levels. Furthermore, PM localization of CAR1 was demonstrated to be related to membrane microdomains. Collectively, our study revealed that CARs recruited the three main components of ABA signaling to the PM to respond positively to ABA. This study deepens our understanding of ABA signal transduction.

biophysics↗

bioRxiv2023

Integrated analysis and annotation for T-cell receptor sequences using TCRosetta

BackgroundT cells and T cell receptors (TCRs) are essential components of the adaptive immune system. TCRs, on the surface of T cells, are responsible for recognizing and binding antigen peptide-MHC complex and play vital roles in T-cell immunology. Characterization of TCR repertoire offers a promising and high informative source for understanding the functions of T cells in immune responses and immunotherapies. Many researchers are now interested in TCR repertoire study; however, there are rare online servers for TCR analysis, especially no servers for TCR annotation and advanced analyses. ResultsWe developed TCRosetta, a comprehensive online server integrating the analytical methods for TCR repertoire/sequences analysis and visualization. TCRosetta combines general features analysis, large-scale sequence clustering, network construction, TCR-peptide binding prediction, generation probability calculation, and k-mer motif analysis for robust TCR sequence analysis, making TCR data analysis as simple as possible and allowing users to concentrate on research rather than coding. In addition, TCRosetta could annotate disease information for TRB CDR3 sequences by batch searching similar sequences in manually curated disease-related TCR database. The TCRosetta server accepts multiple input data formats and can analyze [~]20000 TCR sequences in less than three minutes. ConclusionsTCRosetta is the most comprehensive web server to date for TCR sequences/repertoires analysis and it is freely available at http://bioinfo.life.hust.edu.cn/TCRosetta/. It can be applied to help discover novel biomarkers for disease diagnosis and identify cancer-associated TCR sequences. WHAT IS ALERADY KNOWN ON THIS TOPICT cell receptor repertoires are largely untapped resource than can be used for predicting immune responses to different exposures including viral infections and tumor neoantigens. The downstream analysis of TCR repertoire is often performed by different tools requiring diverse operating environments and expertise. There is no webserver for comprehensive TCR repertoire analysis including general and advanced analysis. WHAT THIS STUDY ADDSWe developed TCRosetta, a comprehensive platform for analyzing T-cell repertoire which combines nearly all TCR analysis methods. It supports different kinds of the format of input including most mainstream TCR extraction tools or amino acid sequences. It could analyze the features of TCR repertoire and display them in interactive graphs and is the first platform with a batch search and TCR annotation function. HOW THIS STUDY MIGHT AFFECT RESEARCH, PACTICE OR POLICYTCRosetta can be applied to discover novel biomarkers to predict response in immunotherapy such TCR repertoire diversity and clonality. It also can identify cancer-associated TCR sequences by clustering biochemically similar CDR3 sequences. It can make TCR repertoire analysis as effortless as possible and help users focus on research instead of coding.

bioinformatics↗

bioRxiv2019

CCLA: an accurate method and web server for cancer cell line authentication using gene expression profiles

Cancer cell lines (CCLs) as important model systems play critical roles in cancer researches. The misidentification and contamination of CCLs are serious problems, leading to unreliable results and waste of resources. Current methods for CCL authentication are mainly based on the CCL-specific genetic polymorphisms, whereas no method is available for CCL authentication using gene expression profiles. Here, we developed a novel method and homonymic web server (CCLA, Cancer Cell Line Authentication, http://bioinfo.life.hust.edu.cn/web/CCLA/) to authenticate 1,291 human CCLs of 28 tissues using gene expression profiles. CCLA curated CCL-specific gene signatures and employed machine learning methods to measure overall similarities and distances between the query sample and each reference CCL. CCLA showed an excellent speed advantage and high accuracy with a top 1 accuracy of 96.58% or 92.15% (top 3 accuracy of 100% or 95.11%) for microarray or RNA-Seq validation data (719 samples, 461 CCLs), respectively. To the best of our knowledge, CCLA is the first approach to authenticate CCLs based on gene expression. Users can freely and conveniently authenticate CCLs using gene expression profiles or NCBI GEO accession on CCLA website.

bioinformatics↗

bioRxiv2019

Systematical analysis reveals the novel function of Cyp2c29 in liver injury

As a severe lethal cancer, hepatocellular carcinoma (HCC) usually originates from chronic liver injury and inflammation, in which the discovery of key genes is important for HCC prevention. Here, we analyzed the time serial (from 0 week to 30 weeks) transcriptome data of liver injury samples in diethylnitrosamine (DEN)-induced HCC mouse model. Through expression and function analyses, we identified that Cyp2c29 was a key gene continuously downregulated during liver injury. Overexpression of Cyp2c29 suppressed the NF-{kappa}B activation, proinflammatory cytokine production and hepatocyte proliferation by increasing its production 14,15-epoxyeicosatrienoic acid (14,15-EET). Furthermore, in vivo Cyp2c29 protected against liver inflammation in liver injury mice models by reversing the expression on functions of cell proliferation, metabolism and inflammation including suppressing NF-{kappa}B pathway and compensatory proliferation. CYP2C8 and CYP2C9, two human homologs of mouse Cyp2c29, were decreased in human HCC progression and positively correlated with HCC patient survival. Therefore, through systematical analysis and verification, we identified that Cyp2c29 is a novel gene in liver injury and inflammation, which may be a potential biomarker for HCC prevention and prognosis.

bioinformatics↗

bioRxiv2019

An ultra-sensitive T-cell receptor detection method for TCR-Seq and RNA-Seq data

T-cell receptors (TCRs) recognizing antigens play vital roles in T-cell immunology. Surveying TCR repertoires by characterizing complementarity-determining region 3 (CDR3) can provide valuable insights into the immune community underlying pathologic conditions, which will benefit neoantigen discovery and cancer immunotherapy. Here we present a novel tool named CATT, which can apply on TCR sequencing (TCR-Seq), RNA-Seq, and single-cell TCR(RNA)-Seq data to characterize CDR3 repertoires. CATT integrated maximum-network-flow based micro-assembly algorithm, data-driven error correction model, and Bayes classification algorithm, to self-adaptively and ultra-sensitively characterize CDR3 repertoires with high accuracy. Benchmark results of datasets from in silico and real conditions demonstrated that CATT showed superior recall and precision compared with other prevalent tools, especially for datasets with short read length and small data size. By applying CATT on a TCR-Seq dataset from aplastic anemia patients, we found the skewing of TCR repertoire was due to the oligoclonal expansion of effector memory T-cells. CATT will be a powerful tool for researchers conducting TCR and immune repertoire studies. CATT is freely available at http://bioinfo.life.hust.edu.cn/CATT.

bioinformatics↗