特邀专家
AI 赋能蛋白质组学大数据挖掘
Abstract:
蛋白质组学大数据具有规模庞大、计算复杂、隐藏信息丰富等特点，传统分析方法难以高效挖掘突变、翻译后修饰等复杂问题，成为制约蛋白质组学成果转化的关键瓶颈。本报告聚焦蛋白质组学大数据挖掘的核心需求，依托人工智能技术与蛋白质组信息学交叉融合优势，系统阐述团队开发的蛋白质组学大数据自动化处理分析生态系统以及 AI 技术在其中的应用路径，探讨 AI 模型如何突破传统分析局限，提升蛋白质组学大数据的挖掘效率与精准度，为推动蛋白质组学与人工智能交叉领域的创新发展提供思路与参考。Keywords:
蛋白质组学；大数据；人工智能；知识挖掘
Dr.AI 数据科学家：海量生物医学数据自主科学分析
Abstract:
随着生物医学研究进入多组学与大数据时代，研发人员正面临前所未有的信息过载与数据处理瓶颈。传统的堆人力模式已难以跟上指数级增长的科研需求。本次报告将介绍“Dr.AI 数据科学家”平台，通过智能体协同构建 AI 自主“读文献—提假设—算数据”的科研闭环（Nature Biomedical Engineering 2026）。Dr.AI 结合文献证据开展链路推理自主提出新的科学假设与研究路径，并面向多模态生物医学数据，自主规划分析流程、调用生物信息工具、编写代码、执行计算并生成可视化结果，形成从知识发现、假设生成到数据验证的端到端科研闭环。Keywords:
AI4S；多智能体
从质谱数据资源到蛋白质组 AI 基础设施：π-MSNet 的构建及其应用
Abstract:
AI 正在快速改变生命科学研究模式，而高质量、大规模、标准化的数据资源已成为推动蛋白质组智能化分析的重要基础。本报告将介绍 π-MSNet——一个面向 AI 应用的十亿级谱图规模蛋白质组学数据资源与分析平台。通过统一元数据标准、自动化重分析流程和高效数据存储格式，π-MSNet 将分散的质谱数据转化为可复用、可扩展、支持模型训练与评估的 AI-ready 资源。基于 π-MSNet，我们进一步探索了其在谱图预测、从头测序、保留时间预测和智能化质谱数据分析中的应用，并展示了大规模蛋白质组数据资源对于计算方法开发和模型性能评估的潜在价值。Keywords:
π-MSNet；AI-ready 数据资源；蛋白质组学人工智能；质谱数据挖掘；深度学习模型
深度学习与镜像酶联用实现高覆盖高可信的肽从头测序
Abstract:
肽从头测序能够鉴定现有数据库之外的蛋白质序列，在抗体、新抗原、突变蛋白、宏蛋白组等测序任务中应用广泛。近年来，尽管基于深度学习的肽从头测序取得了长足进步，但仍面临着物理证据（碎片离子）不足、缺乏质量评估标准等根本问题。本研究将深度学习与镜像酶切技术联用，并引入基于目标—诱饵回贴的质控策略，开发了高覆盖高可信的肽从头测序软件 DiNovo。与单独使用胰蛋白酶相比，使用两对镜像蛋白酶的 DiNovo 获得了两到三倍的高置信度氨基酸序列，并展现出替代传统数据库搜索方法的巨大潜力。Keywords:
肽从头测序；镜像酶；深度学习；质量控制
人工智能驱动的空间组学技术研究
Abstract:
近年来，单细胞及空间组学技术已成为解析组织异质性和复杂细胞相互作用的前沿工具。然而，这类新兴组学数据具有复杂的信息密度和昂贵的实现成本，传统技术手段难以用于全面挖掘解析生物组织复杂组成特征和功能规律。为此，我们专注于“人工智能驱动的空间组学技术研究”，通过深度融合先进的 AI 计算框架，建立了一系列极具创新性的生物信息学解构方法，旨在全面厘清组织的组成和功能异质性。在单细胞信号解构方面，我们建立了胞外小囊泡异质性追踪算法 SEVtras，首次以细胞外尺度描绘了不同类型细胞的生理活动状态；在空间组学方法创新方面，我们开发了针对组织切片的 AI 驱动正交降维采样和重构方法，大幅拓宽了空间多组学测量的技术瓶颈，实现了全组织切片水平的高分辨率空间蛋白质组检测。本次汇报将以全新的纯计算视角，重点分享我们在人工智能驱动空间单细胞组学信息解码领域的思路与突破。我们期望这些工作能够显著提升空间单细胞组学的解析深度与广度，为本领域提供 AI 计算驱动的高性能解析框架，相关成果发表于 Cell, Nature Methods 等国际知名期刊。Keywords:
空间蛋白组学；生物信息学；压缩感知
交联的质谱断裂与数据解析的连接臂架构考量
Abstract:
化学交联质谱中，连接臂的作用常被忽视。但是，交联剂的连接臂不但影响其水溶性，还直接参与断裂，决定后处理流程与解析方法。大肠杆菌裂解液与模式蛋白体系交联实验表明，增减连接臂的亚甲基数目，会直接影响光交联的碎裂通路、报告离子、主链碎片和鉴定结果。从机理角度再认识连接臂的结构与设计，对后续交联剂设计、质谱采集优化、数据解析及原位交联组学研究至关重要。Keywords:
光交联；烷基二氮丙啶；可断裂交联剂；连接臂；奇偶性
Metaproteomic data characteristics and analytical frameworks for microbial ecosystem dynamics
Abstract:
Metaproteomics profiles proteins expressed by microbial communities and links protein-derived functional information to taxonomic contributors, providing a protein-level view of community responses to perturbation. Its data structure combines four interrelated features: the compositional and perturbation-responsive nature of functional profiles, taxon-associated functional profiles, phylogenetic relationships among contributing taxa, and functional redundancy across taxa. Together, these properties reveal how functions are distributed, reorganized, and potentially compensated within microbial communities. This presentation introduces our recent approaches that translate these features into ecological insights. PhyloFunc integrates taxon-resolved functional profiles with phylogeny to quantify community functional distance and state transitions. FRp measures protein-level functional redundancy and its relationship to ecosystem stability and resilience. PhyloPower uses pilot data and analysis-matched ecological distances to estimate statistical power and sample size. Together, these tools advance metaproteomics from protein identification toward quantitative analysis of microbial ecosystem organization and dynamics.Keywords:
metaproteomics; microbial systems ecology; functional redundancy; beta diversity
CNCP-2026 质谱数据评测结果汇报 —— 走进单细胞质谱技术的深水区
Abstract:
随着质谱仪器和数据采集技术的快速发展，单细胞蛋白质组学正向更高的鉴定深度和分析通量迈进。继 2018 年、2020 年和 2023 年三届 CNCP 质谱技术评测之后，本次评测聚焦单细胞蛋白质组学，采用统一制备的人源 HeLa、果蝇 S2 和斑马鱼 PAC2 细胞样品，对多个实验室、不同质谱平台采集的数据进行系统分析。本报告将重点展示不同细胞上样量下的蛋白质鉴定能力，比较不同质谱仪器的技术特点，深入分析平台间性能差异的主要原因，并探讨低上样量条件下的蛋白质修饰鉴定情况，以期客观反映当前单细胞质谱技术的发展水平，为进一步优化实验技术提供参考。Keywords:
蛋白质组学；单细胞；技术评测；蛋白质鉴定
Disease dependent genetic regulation of the plasma proteome in steatotic liver disease
Abstract:
Plasma proteomics offers a powerful window into human physiology and disease, yet robust interpretation requires careful consideration of quantitative accuracy, sample handling, and population-level biological variation. In this talk, I will present results from a population-scale plasma proteomics study of more than 6,000 individuals. I will describe how the plasma proteome is organized into distinct layers of biological variability, shaped by both genetic regulation and physiological processes, and how these features enable accurate prediction of multiple clinical traits. I will also highlight how population proteomics can uncover mechanisms linking genetic variation to liver disease biomarkers and support the development and validation of clinically relevant biomarker panels. Finally, I will share recent observations on the impact of blood collection and sample processing conditions on plasma proteome composition, illustrating how pre-analytical variables influence proteome measurements and affect biomarker discovery and cross-cohort comparability.Keywords:
plasma proteomics; population variation; disease dependent pQTL
An operational perturbation proteomics-based virtual cell model
Abstract:
Artificial intelligence-empowered virtual cell models represent an emerging approach for in silico drug discovery, yet most existing approaches lack large-scale, time-resolved perturbation proteomics data and interpretable frameworks for predicting therapeutic responses. Here we generated more than 38 million temporal protein-abundance measurements from systematically perturbed breast cancer cell lines, and developed ProteinTalks, a virtual cell model. Central to ProteinTalks is the synergy of this large-scale dynamic proteomic resource and the model architecture, enabling a new pretraining framework that learns transferable dynamical latent representations from temporal proteome trajectories. By modelling how proteins respond conditionally to different perturbations, this approach enables the model to function as an operational tool for diverse drug discovery tasks: predicting drug efficacy and synergy, discovering new drug combinations, probing proteins associated with drug resistance, stratifying patient responses and prioritizing drug candidates for patient organoids. It also shows robust transferability, extending beyond cell lines to patient-derived organoids and clinical biopsies, generally achieving higher performance than the selected benchmark implementations under the evaluated protocols. Together, ProteinTalks shows how scalable pretraining of transferable dynamic representations enables operational, dynamics-aware, proteomics-based virtual cell models to advance in silico drug discovery.Keywords:
drug discovery; proteomics; virtual cell
蛋白质通讯的时空化学解析
Abstract:
多细胞生物的功能依赖于细胞间的动态协作，其中生物分子跨细胞转移是关键的通讯机制。细胞间通讯通过多种途径实现，如外泌体、纳米管等，负责传递蛋白质、RNA 等分子甚至是完整细胞器。鉴定这些在不同细胞间传递的生物分子对理解生命活动至关重要，但也面临巨大挑战。传统多组学技术难以获取生物分子的时空动态信息，而新兴的时空分辨多组学也难以有效捕捉这类快速过程。邻近标记技术通过在活细胞内遗传编码酶来记录蛋白质时空动态信息，如 TransitID 技术可解析蛋白质介导的细胞间交流。然而，相关技术在活体应用中常存在高毒性、高背景等问题，严重制约了其在活体动物中探索真实细胞间通讯分子的能力。秦为课题组致力于发展活体兼容的新一代邻近标记技术，用于研究器官之间、细胞之间和细胞器之间等多层次的生物分子交流与通讯机制。Keywords:
蛋白质通讯；化学生物学；邻近标记
An automated deep-learning based de novo-driven pipeline for neoantigen discovery
Abstract:
Mass spectrometry (MS)-based peptidomics has become an indispensable technology for characterizing HLA-presented peptides, but conventional proteogenomic pipelines rely on NGS-derived theoretical peptide databases, which may not fully capture the diverse origins of presented peptides while expanding the search space and reducing identification accuracy and sensitivity. We developed NovoNeo, an automated deep-learning, de novo-driven pipeline for immunopeptide identification, multi-omics approach sourcing, and neoantigen prioritization.We developed a database-free framework to identify neoantigens from diverse sources, including mutations, non-canonical translation, and proteasomal splicing. DeepNovo directly infers peptide sequences from MS/MS spectra, with peptide-level FDR estimated using target-decoy spectra rather than protein databases. Candidate peptides are traced to their sources using matched RNA-seq and prioritized based on tumor-enriched expression using TCGA and GTEx. DeepSelf further predicts HLA binding and immunogenicity to prioritize candidate neoantigens.
Keywords:
de novo sequencing; multi-omics; neoantigen discovery
AI4Proteomics 的发展和普及思考
Abstract:
AI 迅猛发展的今天让大多数人的科研都有了如虎添翼的感觉，虚拟细胞，蛋白模拟等等也都已经占满了我们的朋友圈。本次报告将着眼于本人对近期 AI 发展的一些感受和各个蛋白质组学实验室所见所闻分享对蛋白质组学领域未来一些发展的看法。Keywords:
AI；蛋白质组学进展
两类离子淌度质谱数据的两种计算解析方法
Abstract:
离子淌度质谱通过引入离子淌度维度，为蛋白质组学提供了更丰富的分离信息，同时也对数据解析方法提出了更大的挑战。本报告围绕两类离子淌度质谱数据，介绍两种针对性的计算解析方法。首先，针对基于 diaPASEF 采集的四维数据，介绍以肽为中心的 Full-DIA，通过多维信号提取、深度学习评分及迭代优化，实现高灵敏度的肽段鉴定、准确的蛋白质定量及无缺失值的定量矩阵构建。其次，针对新型无四极杆 SLIM-PAMAF 质谱数据，介绍以谱图为中心的 xTracer，通过离子淌度信号追踪前体—碎片关联，实现非靶向解析。最后，探讨不同采集模式下计算方法面临的挑战与发展方向，为充分挖掘离子淌度质谱数据的信息价值提供新思路。Keywords:
离子淌度质谱；鉴定与定量；计算蛋白质组学
高精度糖蛋白质组学新技术开发和应用
Abstract:
糖蛋白质组的高精度分析依然是生命医学领域的技术挑战之一。我们团队开发的 StrucGP 软件首次实现无需糖链结构数据库即可解析每个糖基化位点上的 N-糖链结构；研发中的 O-糖肽分析软件也正在攻克完整糖肽上成簇 O-糖链的结构解析难题；StrucGAP 分析平台通过八大核心模块实现组学数据的深度挖掘与可视化；预测模型 SpecGP 在糖肽结构图谱预测及鉴定准确性提升方面取得突破。借助上述新技术，系统揭示了 N-糖链的组织特异性分布规律；证实了细胞核 N-糖蛋白的存在及其重要功能；发现了多种组织衰老过程中的糖基化改变模式；并阐明了核心岩藻糖升高调控肝癌上皮间质转化进程的一种新机制。Keywords:
糖蛋白质组学；软件开发；预测模型；衰老研究；肿瘤机制
Streamlined sample preparation workflows for ultrasensitive glycomics of low-input samples
Abstract:
Glycosylation is a critical regulator of biological function and biopharmaceutical efficacy, yet comprehensive glycomic analysis of mass-limited samples remains challenging. Conventional workflows typically require microgram-level protein inputs and often suffer from cumulative sample loss, biased glycan recovery, and glycan degradation during multistep processing, limiting their applicability to trace biological specimens. Moreover, native glycans are highly hydrophilic, which limits their retention and separation by conventional reversed-phase liquid chromatography. To overcome these challenges, we developed streamlined workflows that minimize handling losses and generate hydrophobic derivatives compatible with routine RPLC-MS and MALDI-MS.For N-glycomics, we established solution-enhanced Glycan Reduction and Permethylation (seGRAP), which integrates glycan reduction, desalting, permethylation, and purification within a single PGC microcartridge. Elution with heated DMSO enables a direct transition from PGC desalting to permethylation, while ethyl acetate improves the recovery of permethylated glycans. Coupled with nanoRPLC-MS, seGRAP achieved complete high-mannose N-glycome coverage from 25 pg of RNase B and glycome profiling from 2 pL of human plasma and individual human oocytes. For O-glycomics, we developed One-Pot O-glycan Release and Permethylation (OPORP), which combines nonreductive O-glycan release and in situ permethylation in a single reaction, thereby suppressing peeling reactions and reducing sample handling. OPORP completes sample preparation within ~2 h and enables nanoRPLC-MS profiling from 1 ng of fetuin or 1,000 MCF-7 cells, with CVs below 20% and quantitative linearity of R² values ≥ 0.95.
Together, these streamlined workflows provide accessible platforms for ultrasensitive glycomics of rare cells and other low-input samples.
Keywords:
ultrasensitive glycomics; low-input samples; sample preparation; permethylation
拟南芥 LRR 家族受体激酶的“社交”网络
Abstract:
LRR 家族受体激酶（LRR-RLKs）是拟南芥受体激酶中最大的亚家族，包含 220 余个成员。该类受体激酶定位于细胞膜，在感知环境信号变化中发挥重要功能，调控植物受精、各阶段生长发育以及胞外多种信号的感知与传递等多种生物学过程。LRR-RLKs 可在细胞表面与其他受体形成受体复合体，也可直接调控细胞膜定位的转运蛋白，从而高效、经济地调控离子和多种激素等物质的跨膜运输。本项目以 220 个 LRR-RLKs 为中心，综合运用 AP-MS 和 PL-MS 两种技术，构建了拟南芥膜蛋白互作网络，并从中筛选到 LRR46-SWEET11 模块调控蔗糖长距离运输，为解析蔗糖长距离运输的分子机制提供了重要理论支持。Keywords:
plant membrane PPI network; LRR-RLKs; sucrose transport
面向非模式抗体资源发现的从头测序新方法研究
Abstract:
质谱（MS）技术在抗体序列重建中面临显著挑战，主要源于长肽片段的不完全断裂以及抗体序列固有的高度复杂性，这严重制约了其在非模式生物来源的超长型及非典型抗体分析中的应用。为突破上述瓶颈，本研究开发了一种集成化质谱分析新策略，将微波辅助酸水解（MAAH）与升级版“从头”测序算法 Fusion 2.0 深度融合，系统性提升抗体序列解析能力。MAAH 工作流程可在强酸条件下实现快速、一步式抗体特异性消化，大幅增强长肽段的生成效率与序列覆盖深度。以原始 5 分钟微波处理方案为基准，该方法使长度超过 25 个氨基酸的肽段鉴定数量提升约 25 倍（由 106 个增至 2,645 个）。与此同时，Fusion 2.0 算法在光谱图解析精度、多峰干扰抑制及复杂序列组装鲁棒性等方面均实现显著优化，可稳定识别并准确重构富含重复、高变或低离子化效率的长程结构域。依托该整合平台，我们成功实现了多种结构高度复杂的抗体——包括源自非模式物种的超长型抗体及单域抗体——的深度覆盖“从头”测序；所获序列经重组表达与功能验证，证实其高度准确性与生物学活性。本框架不仅拓展了蛋白质水平抗体测序技术的应用边界，使之可覆盖此前难以表征的天然抗体库，亦为跨物种免疫系统中新型抗体的系统性发现与功能挖掘提供了普适性技术支撑。Keywords:
非模式抗体；质谱；从头测序
基于蛋白质组学的疾病治疗策略研究
Abstract:
本报告围绕蛋白质翻译后修饰与重大疾病发生发展的分子机制，介绍三个领域的研究工作。在肿瘤领域，基于高危前列腺癌临床蛋白基因组学，解析分子分型及代谢重编程，揭示支链氨基酸代谢和 FOXA1 磷酸化驱动肿瘤进展的机制。在感染领域，聚焦分枝杆菌，解析病原菌修饰调控酶介导的宿主磷酸化、乙酰化调控网络及免疫调控机制。在代谢领域，发现全新赖氨酸苯乙酰化修饰，阐明肠道菌群代谢物驱动蛋白修饰并损伤线粒体稳态、促进代谢性肝病的机制。研究形成从临床组学发现到机制解析与靶标发掘的研究体系。Keywords:
蛋白质翻译后修饰；前列腺癌；微生物—宿主互作；新型苯乙酰化修饰
Hybrid DIA MS for scalable proteome digitization and sensitive biomarker monitoring in large cohort studies
Abstract:
Large-cohort and population-scale proteomics require analytical workflows that combine robust, scalable proteome-wide profiling with sensitive and reproducible measurement of predefined biomarkers from limited and heterogeneous biospecimens. Hybrid DIA-MS addresses this need by integrating discovery-driven DIA with targeted PRM acquisition in a single analysis, enabling simultaneous proteome digitization and sensitive biomarker monitoring without requiring separate discovery and targeted workflows. The approach supports both heavy-labeled standard-triggered PRM and adaptive retention time (RT)-enabled PRM with online adjustment of target windows, and was evaluated across Orbitrap Excedion Pro and Orbitrap Astral Zoom platforms. Across translational and plasma proteomics applications, Hybrid DIA-MS improved data completeness and quantitative performance for targeted markers while preserving deep global proteome coverage. On Orbitrap Astral Zoom, the workflow supported up to 999 targeted peptides within a 22-minute, 60-samples/day method while maintaining approximately 9,500 protein groups. Plasma-oriented studies further demonstrated high sensitivity for low-abundance targets, improved quantitative precision, and an approximately one-order-of-magnitude extension in LOQ while retaining broad proteome coverage. By combining scalable proteome-wide digitization with sensitive, hypothesis-driven biomarker measurement in a unified acquisition, Hybrid DIA-MS provides a flexible analytical strategy for large-cohort studies, translational research, and biomarker development.Keywords:
Hybrid DIA MS; large-cohort study; translational research
ChemAcyl libraries enable antibody-free, ultra-low-input profiling of lysine acylation
Abstract:
Lysine acylations, exemplified by lactylation (Klac), regulate diverse biological processes. However, their global profiling typically relies on pan-specific antibodies, which are difficult to generate against diverse chemical moieties, introduce motif-dependent enrichment biases, and require large sample inputs. Here we present an antibody-free framework repurposing acyl-NHS chemistry to convert whole-proteome digests into comprehensive acylation reference libraries. Using Klac as a flagship model, we chemically generated a massive atlas of 250,000 human Klac peptides, which, coupled with a tailored Prosit model, significantly enhanced identifications. This extensibility was validated via lysine acetylation and benzoylation libraries. Leveraging this resource, we developed multiplexed targeted mass spectrometry assays monitoring over 17,000 Klac peptides to extensively map endogenous sites from ultra-low-input samples, including single cells, tumor sections, and primary T cells. By rapidly generating a nicotinylation library for targeted monitoring, we demonstrate that this framework serves as a discovery engine for versatile acylations, bypassing antibody constraints to explore the uncharted acyl-proteome landscape.Keywords:
ChemAcyl；赖氨酸乳酰化；靶向蛋白质组学；平行反应监测；低起始量样本
Identifying peptides with post-translational modifications — from robust identification to combinatorial optimization
Abstract:
Post-translational modification (PTM) plays a critical role in disease biology, yet identifying peptides with multiple PTMs remains computationally challenging. In this talk, we present our effort in tackling this computational challenge by using different methods, including coding method, simulation study, greedy method, and a combinatorial optimization method.Applications to diverse datasets, including lung, colon, and brain cancers, have demonstrated that detection of complex PTM patterns provides new insights into cancer-specific regulatory mechanisms and offers a powerful framework for decoding PTM-driven cellular processes.
Joint work with F. YU, J. DAI, S. LAI, S. DAI, P. ZHAO, C. ZHOU, and N. LI.
