Dissertation defence: DI Johannes Smolander
DI Johannes Smolander defends his dissertation in Computer Science entitled “Cell type identification, differential expression analysis and trajectory inference in single-cell tranomics” at the University of Turku on 31 August 2023 at 12.00 pm (Turku Bioscience Centre, BioCity, Presidentti auditorium, Tykistökatu 6, Turku).
Opponent: Professor Mark Robinson (University of Zurich, Switzerland)
Custos: Professor Laura Elo (University of Turku)
Summary of the Doctoral Dissertation
Single-cell RNA-sequencing (scRNA-seq) is a cutting-edge technology that enables to quantify the transcriptome, the set of expressed RNA transcripts, of a group of cells at the single-cell level. It represents a significant upgrade from bulk RNA-seq, which measures the combined signal of thousands of cells. Measuring gene expression by bulk RNA-seq is an invaluable tool for biomedical researchers who want to understand how cells alter their gene expression due to an illness, differentiation, ternal stimulus, or other events. Similarly, scRNA-seq has become an essential method for biomedical researchers, and it has brought several new applications previously unavailable with bulk RNA-seq.
scRNA-seq has the same applications as bulk RNA-seq. However, the single-cell resolution also enables cell annotation based on gene markers of clusters, that is, cell populations that have been identified based on machine learning to be, on average, dissimilar at the transcriptomic level. Researchers can use the cell clusters to detect cell-type-specific gene expression changes between conditions such as case and control groups. Clustering can sometimes even discover entirely new cell types. Besides the cluster-level representation, the single-cell resolution also enables to model cells as a trajectory, representing how the cells are related at the cell level and what is the dynamic differentiation process that the cells undergo in a tissue.
This thesis introduces new computational methods for cell type identification and trajectory inference from scRNA-seq data. A new cell type identification method (ILoReg) was proposed, which enables high-resolution clustering of cells into populations with subtle transcriptomic differences. In addition, two new trajectory inference methods were developed: scShaper, which is an accurate and robust method for inferring linear trajectories; and Totem, which is a user-friendly and flexible method for inferring tree-shaped trajectories. In addition, one of the works benchmarked methods for detecting cell-type-specific differential states from scRNA-seq data with multiple subjects per comparison group, requiring tailored methods to confront false discoveries.
Doctoral Dissertation at UTUPub: https://www.utupub.fi/handle/10024/175558