Multi-Disease Characterization and Classification through Transcriptomics–Based Multi-Omics Analysis
Time: Wed 2026-06-10 13.00
Location: Air and Fire, SciLifeLab, Tomtebodavägen 23
Video link: https://kth-se.zoom.us/j/67142088168
Language: English
Subject area: Biotechnology
Doctoral student: Mengzhen Li , Systembiologi, Science for Life Laboratory, SciLifeLab
Opponent: Professor Ming Chen, Zhejiang University, Kina
Supervisor: Professor Adil Mardinoglu, Systembiologi, Science for Life Laboratory, SciLifeLab, King's College London; Senior Lecturer Cheng Zhang, Science for Life Laboratory, SciLifeLab, Systembiologi, King's College London
QC 2026-05-18
Abstract
Transcriptomics acts as a bridge between genomics and phenotypic outcomes, serving as a powerful tool for understanding system-level changes in human biology due to its high coverage, dynamic regulatory capacity, and ability to capture early responses. This thesis focuses on characterizing transcriptomic signatures across different tissues, diseases, and further integrates proteomic and metabolic data to provide a more comprehensive view of diverse health conditions.
In Papers I and II, we applied transcriptomic analyses to cell line models to characterize the features of an in vitro steatosis model and to investigate the effects of herbal treatment on cancer cell lines. In Paper III, we explored whether whole blood transcriptomics (WBT) can capture disease-associated features and developed a WBT-based disease prediction pipeline, addressing challenges related to cohort heterogeneity and transcriptomic platform batch effects. In Paper IV, we established a large-scale cohort comprising 4,444 participants across 98 health conditions and constructed a comprehensive Whole Blood Transcriptomic Atlas (WBT Atlas). We systematically characterized disease-specific features from immunological and metabolic perspectives at the gene, gene module, and pathway levels. In addition, we integrated matched proteomic data to highlight the complementary nature of transcriptomics and proteomics. Finally, we developed multi-disease detection models that enable individual disease risk classification. In Paper V, we extended transcriptomic analysis to the metabolic layer by constructing 32 tissue-specific and 81 cell-type-specific enzyme-constrained genome-scale metabolic models (ecGEMs) across the human body, linking gene expression to metabolic function.
In summary, this thesis adopts a transcriptomics-centered framework for multi-omics integration across diverse tissues and diseases, providing insights into disease mechanisms and enabling biomarker-based disease prediction.