Extracting information from gene expression data using computational methods

Differences in the gene expression of samples from healthy and sick donors can be applied for many purposes, such as predicting disease onset, identifying disease subtypes, understanding the processes related to disease progression, or selecting personalized treatment. In her thesis, Maria Jaakkola evaluated and developed computational methods to extract information from large gene expression datasets.

Gene expression data can be used to estimate the activities of different signalling pathways. Signalling pathways are interaction networks that body uses to transfer information and react to changing situations. In the thesis, different pathway methods are tested with diverse datasets and methods utilising pathway structure are shown to outperform simple over-representation based tests. A new method estimating pathway activities of individual samples rather than sample groups is also presented, which is important for applications in personalised medicine.

However, detecting differences between the sample groups is difficult if only one cell type is altered and noise from the other present cell types masks the signal.

– This is an issue as many clinically easy samples, like blood, contain several cell types, Jaakkola says.

In the thesis, different methods to separate the expression originating from different cell types are evaluated and a novel method particularly suitable for difficult data with high variability within sample groups is introduced. In addition, practical guidelines how the end-user can evaluate if reliable results can be obtained from their particular dataset were provided.

***

MSc Maria Jaakkola presents her thesis ”EXTRACTING INFORMATION FROM HIGH-THROUGHPUT GENE EXPRESSION DATA WITH PATHWAY ANALYSIS AND DECONVOLUTION” for public examination in the University of Turku on Friday May 6th 2022 at 12.00 (University of Turku, Biocity, Ministeri- auditorium, Tykistökatu 6, Turku).

Professor Sorin Draghici (Wayne State University, United States) has been nominated as the opponent and Professor Laura Elo (University of Turku) as the custos. The field of study is applied mathematics and the language of the dissertation will be English.

The thesis has been published online: https://urn.fi/URN:ISBN:978-951-29-8846-4

The University of Turku actively follows the coronavirus situation and the authorities’ instructions. The University may change the instructions according to the situation.