Full text
| |
| Author(s): |
Tiago Lubiana Alves
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | São Paulo. |
| Institution: | Universidade de São Paulo (USP). Instituto de Matemática e Estatística (IME/SBI) |
| Defense date: | 2020-03-31 |
| Examining board members: |
Helder Takashi Imoto Nakaya;
André Fujita;
Israel Tojal da Silva;
Eduardo Lani Volpe da Silveira
|
| Advisor: | Helder Takashi Imoto Nakaya |
| Abstract | |
The advent of single-cell RNA sequencing has brought scientific advances and technical challenges. The development of new analysis methodologies is a crucial step to maximize the extraction of knowledge from this data. In this work, we explore the repositioning of a feature selection algorithm to address the analytical challenges of public single-cell RNA data. We adapted the FCBF method(Fast Correlation-Based Filter) to select relevant genes to distinguish cell types, and, from this list of genes, find modules of coexpressed genes. In blood cell data, we noted that the modules found corresponded to transcriptional programs characteristic of specific cell groups. As a result, we implemented a pipeline capable of using the coexpression modules to infer the cell types present in datasets in a multilevel way, avoiding the limits of traditional single labels. We processed datasets from human blood mononuclear cells and zebrafish embryonic cells, noting that the modules and populations found brought to light biologically relevant information. The algorithms were implemented in two Bioconductor packages, FCBF and fcoex, and are available to the community,increasing the arsenal for analyzing single-cell RNA sequencing data. (AU) | |
| FAPESP's process: | 18/10257-2 - Neuroimmune molecular networks in neurodegenerative and in infectious diseases |
| Grantee: | Tiago Lubiana Alves |
| Support Opportunities: | Scholarships in Brazil - Master |