Development of web-based tools for metabolomics data analysis and mining
Year: 2017
Type: dissertation
Abstract: The recent advances in metabolomics experimental techniques have provided novel approaches for many research issues in the biological fields. Indeed, the ability to identify and quantify numerous compounds in biological samples provides significant advances in functional genomics, biomarker identification, sample characterization or drug discovery and development. To take full advantage of these data advanced bioinformatics methods for data analysis and mining have been required. A number of methods and tools for metabolomics data analysis have been put forward recently, being one of the major limitations still faced the lack of integrated frameworks for extracting relevant knowledge from these data and being able to integrate these data with previous biochemical knowledge. Also, the lack of reproducibility in many data analyses or data mining processes is a strong obstacle for biological discovery. In recent work from the host group, specmine, a metabolomics and spectral data analysis/ mining framework, in the form of a package for the R system, has been developed to address some of these issues. In this thesis, an integrated web-based platform for metabolomics data analysis and mining, named WebSpecmine, was designed and developed, based on the specmine package, thus providing an easier and friendly user interface. This website provides means for analysing metabolomics data from different formats, including tasks such as pre-processing, univariate and multivariate analysis and metabolite identification. This web-based platform was developed collaboratively and, therefore, this work focused mainly in data from nuclear magnetic ressonance and mass spectrometry. Also, the package faced some limitations regarding types of analysis not yet provided, such as metabolite identification for other data formats besides Mass Spectrometry coupled to Liquid Chromatography. Therefore, the extension of the metabolite identification feature was addressed, by implementing such analysis for Nuclear Magnetic Ressonance data in the specmine package, as well as making it available in the website. The website was validated by applying it to reproduce the pipelines from previous studies that made use of the specmine package. Furthermore, a case study involving banana peels and the analysis of their characteristics and potential made use of the newly created website to further validate its functionality. All the analyses here executed were stored and are available in the web application, as public projects.
Author: Sara Cardoso
Institution: University of Minho
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W2890944241
