|
|
Bioinformatic analysis of mass spectrometry data in metabolomics
Skoryk, Maksym ; Raček,, Tomáš (oponent) ; Mgr. Aleš Křenek, Ph.D (vedoucí práce)
This Master's thesis explores and compares methods for the analysis of mass spectrometry data, focusing on the construction and application of molecular networks. The primary objective of this study is to identify suitable mass spectra similarity metrics for the construction of molecular networks, which would reveal meaningful relationships between compounds and their structural and biological properties. To achieve this goal, we thoroughly investigated the performance of different mass spectra similarity metrics, including cosine similarity, spectral entropy, as well as machine learning-based Spec2Vec, and MS2DeepScore. We then applied dimensionality reduction techniques, such as t-SNE, UMAP, PHATE, and Isomap, to visualize and to better understand the molecular networks generated from these metrics. Our results demonstrate the importance of selecting appropriate similarity metrics and their adjustment for particular datasets and usecases. This thesis contributes to the field of untargeted mass spectrometry and metabolomics by investigating the applications of molecular networking to electron ionization gas chromatography-mass spectrometry data.
|
host ::
RSS.