At HUPO 2025, the HUPO Education and Training Committee (ETC) delivered three pre-congress training sessions, each organized and chaired by internationally recognized experts and designed to address different but complementary aspects of modern proteomics.
Session1: Exploring Protein Post-Translational Modifications (PTMs)
This session introduced contemporary strategies for studying protein post-translational modifications, with an emphasis on practical workflows, clinical sample preparation, and emerging directions in PTM and glycoproteomics research. Dr. Hu Zhou (Shanghai Institute of Materia Medica, CAS, China) provides in-depth guidance on PTMomics sample preparation, with a strong focus on practical considerations and best practices for clinical tumor samples. Dr. Nick Riley (University of Washington, USA) walks participants through a comprehensive glycoproteomics workflow and outlines emerging directions in glycoproteomics research. Together, the session highlights modern mass spectrometry–based strategies, key technical challenges, and future opportunities in PTM-focused proteomics, equipping students with actionable knowledge and a broader perspective on PTM biology and disease mechanisms. The session was chaired by Dr. Shixia Huang (Baylor College of Medicine) and Dr. Yansheng Liu (Yale University).
Session 2: Mastering Computational Proteomics
The three talks collectively emphasized that although proteomics now produces vast amounts of LC–MS/MS data, its full potential—particularly for experiment planning and AI-driven discovery—remains restricted by inconsistent metadata and technical variability across datasets. Dr. Deepti Jaiswal Kundu (EMBL-EBI) highlighted how standards such as the Sample and Data Relationship Format (SDRF) enable datasets to become machine-readable and reproducible, helping transform repositories like PRIDE and MassIVE from passive archives into genuinely reusable scientific resources. Dr. Tine Claeys (VIB-UGent Center for Medical Biotechnology) reinforced this message from an AI perspective, showing that meaningful, consistent annotation is essential for training reliable models such as MLMarker, and revealing how missing information, batch effects, and format inconsistencies silently undermine large-scale data reuse. Dr. Brian Searle (Mayo Clinic, USA) then demonstrated how deep learning tools such as Prosit can harmonize fragmented datasets into a shared coordinate framework, enabling the training of more generalizable models and unlocking the value of legacy experiments. Together, the talks made a compelling case that rigorous metadata standards and data harmonization are fundamental to robust data reuse and biologically meaningful AI in proteomics. The session was chaired by Dr. Deepti Jaiswal Kundu (EMBL-EBI, UK) and Dr. Giuseppe Palmisano (University of São Paulo, Brazil)
Session 3: Single-Cell Proteomics - Emerging Technologies and Translational Applications
The three talks collectively emphasized that single-cell proteomics is rapidly becoming a practical engine for resolving cellular heterogeneity and enabling precision medicine, driven by parallel advances in sample preparation, mass spectrometry, and computational analysis. Dr. Claudia Ctortecka (Broad Institute, USA) underscored that “precision starts with preparation,” walking through experimental design, randomization, QC, and workflow automation—highlighting ultra-sensitive acquisition strategies (including timsTOF Ultra2) that can quantify thousands of proteins per single cell and extend to low-cell-count tissue regions. Dr. Olga Vitek (Northeastern University, USA) reinforced that translation depends on robust statistics: she dissected how biological/technical variation, unbalanced designs, and uncertain cell-type labels can mislead differential analysis and provided practical recommendations for cell-type annotation and modeling choices that improve accuracy. Dr. Ana Konvalinka (University Health Network/University of Toronto, Canada) then connected these capabilities to clinical impact, illustrating how single-cell, tissue, and spatial proteomics can illuminate immune–parenchymal interactions, explain disease heterogeneity, and inform precision monitoring and future trial design. Together, the session made a clear case that scalable, reproducible end-to-end workflows spanning careful sample handling, appropriate statistical inference, and clinically grounded interpretation are essential for bringing single-cell proteomics into real-world translational and clinical use. The session was co-chaired by Dr. Ana Konvalinka (University Health Network/University of Toronto, Canada) and Dr. Lin Qingsong (National University of Singapore).
Summary of Participant Survey Results
Overall Summary
A total of 46 participants completed the post-congress survey across the three HUPO 2025 ETC pre-congress training sessions. Overall feedback was highly positive, with the large majority of respondents indicating that they would recommend the sessions to colleagues (40/46 Agree or Strongly Agree) and that they were interested in attending future HUPO ETC training events (40/46 Agree or Strongly Agree). Participants consistently valued the balance of conceptual grounding and practical insight, particularly where sessions addressed real-world challenges in experimental design, data interpretation, and emerging technologies. Qualitative comments emphasized the importance of expert-led instruction, clear workflows, and training that directly supports reproducibility and data reuse in proteomics research.
Session-Specific Survey Results
Exploring Protein Post-Translational Modifications (PTMs):
12 respondents completed the survey for this session. Feedback was uniformly positive, with respondents highlighting the practical guidance on PTMomics sample preparation, especially for clinical tumor specimens, and the clear, end-to-end overview of glycoproteomics workflows. Participants particularly valued the combination of hands-on considerations with forward-looking perspectives on PTM biology and disease mechanisms. Interest in future HUPO ETC training was very strong among this group, reflecting demand for structured, student-focused PTM education.
Mastering Computational Proteomics:
A total of 26 respondents identified this session (including minor duplicate session labels). Participants consistently emphasized the importance of metadata standards, data harmonization, and reproducibility as key take-home messages. Respondents noted that the session effectively clarified why inconsistent annotation and technical variability limit large-scale data reuse and AI-driven discovery, and they appreciated concrete examples involving SDRF, PRIDE/MassIVE, ML-based modeling, and deep-learning tools such as Prosit. The session was viewed as particularly valuable for trainees planning to work with large or legacy proteomics datasets.
Single-Cell Proteomics: Technologies and Applications:
7 respondents completed the survey for this session. Feedback indicated strong appreciation for the session’s clear evaluation of current single-cell proteomics technologies, including strengths, limitations, and appropriate use cases. Participants valued the balanced discussion of technical performance, data quality considerations, and biological interpretation, which helped them critically assess when and how single-cell proteomics approaches can be effectively applied. Several respondents noted that the session improved their ability to evaluate experimental design choices and technology readiness, rather than simply learning tool-specific workflows. Overall satisfaction was high, and respondents expressed interest in additional ETC offerings that similarly emphasize critical evaluation and practical decision-making in emerging proteomics technologies.
.png)
