Bibliographic Details
| Title: |
Effect of single-cell RNA sequencing parameters on cluster stability |
| Authors: |
Potgieter, Tiqvah Talya |
| Contributors: |
Maasdorp, Elizna, Patterton, Hugh, Stellenbosch University. Faculty of Science. Center for Bioinformatics and Computational Biology. |
| Publisher Information: |
Stellenbosch University |
| Publication Year: |
2025 |
| Collection: |
Stellenbosch University: SUNScholar Research Repository |
| Subject Terms: |
Gene expression -- Data processing, Cluster analysis -- Data processing, Stability, Nucleotide sequence -- Technique, Cells -- Analysis, Single-cell RNA sequencing, UCTD |
| Description: |
Thesis (MSc)--Stellenbosch University, 2025. ; Potgieter, T. T. 2025. Effect of Single-Cell RNA Sequencing Parameters on Cluster Stability. Unpublished masters thesis. Stellenbosch: Stellenbosch University [online]. Available: https://scholar.sun.ac.za/items/7220d82f-7182-4eb7-8dfd-90bd03c01c16 ; ENGLISH ABSTRACT: Single-cell RNA sequencing has allowed us to study gene expression at a single-cell level, enabling researchers to unravel the heterogeneity and complexity within individual cells. Clustering has made this possible as it identifies and groups different cells based on their expression profiles, clustering also plays a critical role in exploratory research and thus, understanding the quality of clustering and clustering outcomes is essential. However, cluster analysis remains one of the most challenging and fundamental tasks in a scRNA-seq analysis. This is mainly due to the fact that there is no ground truth or gold standard by which performance can be measured. Additionally, clustering itself is governed by a range of parameters and processing methods. Cluster stability assessments aim to address these challenges by defining how robust clustering results are under perturbations to the original dataset. The assumption is that stable clustering results represent the underlying structure of the data and thus should be consistent across an ensemble of perturbed datasets. We investigated the effects of various processing parameters on cluster stability in scRNA-seq data. For each parameter, cluster stability was quantified by generating a cluster ensemble with respect to different initializations and calculating the average scores using relevant stability metrics. Results presented here showed great variability, indicating that cluster stability is sensitive to the choices of parameter setting as well as the type of dataset. We emphasize that users should not rely on a once-off application of a clustering algorithm to their data but explore various parameter possibilities and assess the robustness of ... |
| Document Type: |
thesis |
| File Description: |
xi, 90 pages : illustrations; application/pdf |
| Language: |
English |
| Relation: |
https://scholar.sun.ac.za/handle/10019.1/132494 |
| Availability: |
https://scholar.sun.ac.za/handle/10019.1/132494 |
| Rights: |
Stellenbosch University |
| Accession Number: |
edsbas.17575DAA |
| Database: |
BASE |