Medical Biology - Theses
Permanent URI for this collection
Search Results
1 - 1 of 1
-
ItemStatistical and Machine Learning models for estimation of missing values in label-free mass spectrometry quantification( 2022)Mass spectrometry (MS) enables high throughput identification and quantification of proteins in complex biological samples and can provide insights into the global function of biological systems, aberrations and disease progression. Label-free quantification is cost effective, suitable for analysis of human samples and can profile proteins from a broad range of abundance. Despite rapid developments in label-free data acquisition workflows, the number of proteins commonly quantified across samples can be limited. This results in missing values in the measurements between samples, which present substantial challenges for downstream data analysis tasks and biomedical discoveries. This thesis provides two solutions for the treatment of missing values in label- free mass spectrometry: (i) imputation of missing values after quantification using Barycenter computation from Optimal Transport discipline in Machine Learning research, and (ii) a deep learning solution for sequence identification transfer between precursor ions across samples at the quantification step. In the two methodological manuscripts arose from this thesis, I demonstrate how these two solutions enhance data completeness in label-free mass spectrometry acquisition, thereby facilitating biomedical discoveries. I then provide a perspective on the future directions of these two works. The tools developed in this work are available on open-source software repositories and are used by the proteomics and bioinformatics community in medical research projects.