Professor Arndt von Haeseler, Center for Integrative Bioinformatics Vienna (CIBIV), University of Vienna and Medical University of Vienna, Austria

Title: FloVelo: Pushing Boundaries of RNA Velocity Models in scRNA-seq Data

Abstract:

Single-cell RNA sequencing (scRNA-seq) provides impressive new insights into the developmental states of single cells within a biological sample. Recently, numerous algorithms have been introduced to model the dynamics of scRNA-seq based cell differentiation. One of these approaches, RNA velocity, utilizes the knowledge that unspliced mRNA is eventually processed to spliced mRNA and predicts future cell expression profiles based on changing ratios of unspliced to spliced molecules. While current methods compete in improving the prediction of RNA velocity based global developmental trends, we show that the ratio of unspliced to spliced transcripts already provides a deeper understanding of a gene’s splicing dynamic.

To this aim, we present FloVelo, a computational approach, that for each gene determines the most likely cell path in the space of unspliced vs spliced mRNA expression levels. Interpreting the distribution of cells as a probability measure, such paths appear as a one-dimensional density ridge that FloVelo reconstructs by solving multiple, interconnected network flow problems. As a model-free approach, FloVelo is able to resolve complex transcriptional contexts such as cell-state and lineage specific splicing dynamics or transcriptional bursts.

Finally, detected paths are provided in comprehensive, gene-wise output plots and subsequently, path shapes are compared across genes. This provides an intuitive way to classify genes potentially underlying similar regulatory mechanisms, as we can show, for example, in the context of cycling progenitor cells: Distinguishing between cyclic and non-cyclic path shapes identifies cell cycle-related and cell cycle-independent genes. Moreover, FloVelo has the potential to uncover different splicing dynamics of a gene across multiple samples without the need and thus well-known pitfalls of prior scRNA-seq sample integration. FloVelo is a data-driven, explorative way to analyze a new information layer in scRNA-seq data and is therefore able to efficiently deal with future technological developments, such as improved estimation of ratios of unspliced reads and splice variants as made possible by long read sequencing technologies.

Bio:

N/A

Professor Thomas Ågotnes, Head of the Logic and AI research group at the Department of Information Science and Media Studies, University of Bergen, Norway

Title: Group Knowledge: Some Recent Developments

Abstract:

What we mean when we say that a group knows something can be radically different depending on context. Well-known notions of group knowledge that have been proposed in the literature include distributed knowledge (that which follows from the combined knowledge in the group), general knowledge (everybody-knows), common knowledge (everybody-knows that everybody-knows and so on). Although these concepts have been well-known and used for a long time, there has been considerable development in our understanding of them in recent years. This holds in particular for distributed knowledge in general, the relationship between distributed knowledge as potential knowledge and common knowledge as actualised knowledge, the dynamics of distributed knowledge, group knowledge for weak notions of knowledge, and stronger variants of distributed knowledge such as “somebody-knows”. In the talk I will report on some of these new developments.

Bio:

Thomas Ågotnes is a Professor of Information Science and head of the Logic and AI research group at the Department of Information Science and Media Studies, University of Bergen, Norway. He is also a visiting Professor at the Centre for the study of Logic and Intelligence at Southwest University, China. His main research interests include formal knowledge representation and reasoning about different types of interaction, in particular using modal logic, often combined with other mathematical models of interaction for example from game theory, with applications in the fields of artificial intelligence and multi-agent systems. He has published extensively in these fields. He has been awarded the Changjiang (Yangtze river) Scholar award by the Chinese government, and best paper awards at conferences such as AAMAS and CLAR.

N/A