| Popis: |
A trajectory in physics is the path that an object covers in space and time. In analogy, trajectories in biology can be understood as paths that biological entities such as cells or organisms cover between developmental or evolutionary states in space and time. Trajectory inference methods have been developed to analyze molecular ‘snapshot’ data obtained from cross-sectional OMICs experiments to deduce a proxy of temporal processes like cellular development or evolution of an organism that is otherwise often not available. In this thesis, the concept of OMICs-based trajectories was addressed in the context of three topics: (i) Trajectories in single-cell or bulk transcriptomic state space to infer the development and differentiation of cells in a multicellular organism (planaria) as an example for differentiation from stem cells towards differentiated tissues and of the development of treatment resistance of melanoma cell lines under targeted treatment. Further, we analyze the aging of the blood transcriptome in a large population cohort. (ii) Transferring the concept of trajectories from cell state space into gene state space to infer the switching of genomic programs during development. (iii) Transferring the concept of trajectories from transcriptome state space to genetic (mutational) state-space to infer evolutionary paths with applications to SARS-COV2 virus evolution over two years in 2020 and 2021, and dissemination of vine-varieties over about 8,000 years from the (presumably) first cultivation of vine in the Caucasus region around the Mediterranean Sea and Western Europe. The main aim of this thesis is to develop, adapt, and apply computational methods to reconstruct developmental trajectories across different OMICs domains in cell and in feature space. The latter one, trajectories in feature state-space, can be modeled using self-organizing maps (SOM) machine learning. The method transforms multidimensional gene expression patterns into two-dimensional data landscapes resembling the metaphoric Waddington epigenetic landscape. Trajectories in this SOM landscape visualize transcriptional programs passed by cells along their line of development from undifferentiated to differentiated cells and evolutionary paths over several generations of species or cultivars. As a proof-of-principle study, we applied our method to a single-cell transcriptomics data set describing tissue development in a regenerating flatworm. Using SOM machine learning, we identified cell type and tissue-specific modules of activated genes, as well as differentiation trajectories linking stemness-source and tissue-related sink modules. Our methods developed in this thesis enable a novel, more comprehensive view on the dynamics in high-dimensional OMICs data landscapes. The variety of examples chosen illustrates the mutual dependencies as well as the particularities of each state and data type. Our analyses support the understanding of the dynamics of the data in the context of their developmental, functional, geographical, or historical stage. By applying trajectory inference methods to cancer and SARS-CoV-2 data sets, we could confirm and supplement previous findings on the dynamics of important underlying transcriptional and genomic programs.:Table of Contents Abstract 5 1 Introduction 9 1.1 Developmental trajectories 9 1.2 Measuring gene and cell States: challenges in high-throughput data analysis 9 1.3 Single-cell RNA sequencing (scRNA-seq) data and trajectory inference 10 1.4 Trajectories in cell and gene states 13 1.5 Single-cell self-organizing map - portrayal 13 1.6 Objectives and outline 15 1.7 Biological background of OMICs terminologies 16 1.8 Bulk OMICs data 17 1.9 Single-cell OMICs 19 2 Bioinformatics of single-cell RNA sequencing 19 2.1 Processing steps 19 2.2 Single-cell SOM portrayal 23 3 Methods: Pseudotime trajectories in gene and cell state space 25 3.1 Trajectories in cell state space 25 3.2 Trajectories in gene state space 27 3.3 RNA Velocity in gene and cell state space 27 3.4 Software 28 3.5 Application in case studies 28 3.6 Worked example: Trajectories of planarian tissue development 29 4 Application I: Trajectories of melanoma treatment resistance 44 4.1 Results 44 4.2 Discussion 54 4.3 Conclusion 57 5 Application II: Trajectories of the aging blood 59 5.1 Results 59 5.2 Discussion 75 5.3 Conclusion 80 6 Application III: Trajectories of the evolving SARS-CoV-2 genome 81 6.1 Results 82 6.2 Discussion 101 6.3 Conclusion 105 7 Application IV: Trajectories of grapevine dissemination 106 7.1 Results 106 7.2 Discussion 115 7.3 Conclusion 119 8 General Discussion 120 8.1 Trajectory inference across OMICs modalities and timescales 120 8.2 Challenges of trajectory inference of bulk and genomic data 121 8.3 Trajectory inference in cell and gene state space 122 9 Conclusion 125 10 Supplement 126 10.1 Materials and methods 126 10.2 Results 138 Index of abbreviations 207 List of tables 211 List of figures 212 References 215 Curriculum vitae 233 List of publications 234 Conferences / summer schools 236 Selbstständigkeitserklärung 237 |
