No description available.

The self-organizing maps portraying has been proven to be a powerful approach for analysis of transcriptomic, genomic, epigenetic, single-cell, and pathway-level data as well as for “multi-omic” integrative analyses. However, the SOM method has a major disadvantage: it requires the retraining of the entire dataset once a new sample is added, which can be resource- and timedemanding. It also shifts the gene landscape, thus complicating the interpretation and comparison of results. To overcome this issue, we have developed two approaches of transfer learning that allow for extending SOM space with new samples, meanwhile preserving its intrinsic structure. The extension SOM (exSOM) approach is based on adding secondary data to the existing SOM space by “meta-gene adaptation”, while supervised SOM portrayal (supSOM) adds support vector machine regression model on top of the original SOM algorithm to “predict” the portrait of a new sample. Both methods have been shown to accurately combine existing and new data. With simulated data, exSOM outperforms supSOM for accuracy, while supSOM significantly reduces the computing time and outperforms exSOM for this parameter. Analysis of real datasets demonstrated the validity of the projection methods with independent datasets mapped on existing SOM space. Moreover, both methods well handle the projection of samples with new characteristics that were not present in training datasets.

Einzelzell-Omics-Daten stehen derzeit im Fokus der Entwicklung computergestützter Methoden in der Molekularbiologie und Genetik. Einzelzellexperimenten lieferen dünnbesetzte, hochdimensionale Daten über zehntausende Gene oder hunderttausende regulatorische Regionen in zehntausenden Zellen. Diese Daten bieten den Forschenden die Möglichkeit, Gene und regulatorische Regionen zu identifizieren, welche die Bestimmung und Aufrechterhaltung der Zellidentität koordinieren. Die gängigste Strategie zur Identifizierung von Zellidentitätsmarkern besteht darin, die Zellen zu clustern und dann Merkmale zu finden, welche die Cluster unterscheiden, wobei davon ausgegangen wird, dass die Zellen innerhalb eines Clusters die gleiche Identität haben. Diese Annahme ist jedoch nicht immer zutreffend, insbesondere nicht für Entwicklungsdaten bei denen sich die Zellen in einem Kontinuum befinden und die Definition von Clustergrenzen biologisch gesehen potenziell willkürlich ist. Daher befasst sich diese Dissertation mit Clustering-unabhängigen Strategien zur Identifizierung von Markern aus Einzelzell-Omics-Daten. Der wichtigste Beitrag dieser Dissertation ist SEMITONES, eine auf linearer Regression basierende Methode zur Identifizierung von Markern. SEMITONES identifiziert (Gruppen von) Markern aus verschiedenen Arten von Einzelzell-Omics-Daten, identifiziert neue Marker und übertrifft bestehende Marker-Identifizierungsansätze. Außerdem ermöglicht die Identifizierung von regulatorischen Markerregionen durch SEMITONES neue Hypothesen über die Regulierung der Genexpression während dem Erwerb der Zellidentität. Schließlich beschreibt die Dissertation einen Ansatz zur Identifizierung neuer Markergene für sehr ähnliche, dennoch underschiedliche neurale Vorlauferzellen im zentralen Nervensystem von Drosphila melanogaster. Ingesamt zeigt die Dissertation, wie Cluster-unabhängige Ansätze zur Aufklärung bisher uncharakterisierter biologischer Phänome aus Einzelzell-Omics-Daten beitragen. / Single-cell omics approaches are the current frontier of computational method development in molecular biology and genetics. A single single-cell experiment provides sparse, high-dimensional data on tens of thousands of genes or hundreds of thousands of regulatory regions (i.e. features) in tens of thousands of cells (i.e. samples). This data provides researchers with an unprecedented opportunity to identify those genes and regulatory regions that determine and coordinate cell identity acquisition and maintenance. The most common strategy for identifying cell identity markers consists of clustering the cells and then identifying differential features between these clusters, assuming that cells within a cluster share the same identity. This assumption is, however, not guaranteed to hold, particularly for developmental data where cells lie along a continuum and inferring cluster boundaries becomes non-trivial and potentially biologically arbitrary. In response, this thesis presents clustering-independent strategies for marker feature identification from single-cell omics data. The primary contribution of this thesis is a linear regression-based method for marker feature identification from single-cell omics data called SEMITONES. SEMITONES can identify markers or marker sets from diverse single-cell omics data types, identifies novel markers, outperforms existing marker identification approaches. The thesis also describes how the identification of marker regulatory regions by SEMITONES enables the generation of novel hypotheses regarding gene regulation during cell identity acquisition. Lastly, the thesis describes the clustering-independent identification of novel marker genes for highly similar yet distinct neural progenitor cells in the Drosophila melanogaster central nervous system. Altogether, the thesis demonstrates how clustering-independent approaches aid the elucidation of yet uncharacterised biological patterns from single cell-omics data.

No description available.