September 1
Speaker: Calina Copos (Northeastern University)
Title: Emergence of biased behavior in small cell clusters
Abstract: Left/right asymmetry is a hallmark of living systems, where local interactions among energy-consuming components generate organized large-scale organization. To isolate the minimal ingredients underlying this process, we study cell doublets as the smallest nontrivial collective unit.
Combining experiments and mechanobiochemical modeling, we show how mechanical coupling between cells can generate (and modulate) a directionality bias in collective migration. The first example focuses on a mechanochemical model of cell pairs migrating on flat substrates, where we have identified two motility modes: the individual contributor (IC) mode, in which each cell generates its own traction force dipole, and the supracellular (S) mode, characterized by a single dipole spanning the pair. These motility modes are observed in experimental systems, but with different biases depending on the cell type — amoeboid cells predominantly adopt the IC mode, while mesenchymal cells favor the S mode. We show that the model can capture the emergence of both motility modes and that importantly, not cell-cell interactions, but the extent of protrusions modulate which motility mode is preferred. Lastly, we will discuss a macroscopic, agent-based model for bias in the rotational movement of cell pairs. Together, these models illustrate how symmetry breaking, local interactions, and environmental geometry combine to produce directed behavior in cellular systems.
September 8
Speaker: Yuta Hozumi (Georgia Tech)
Title: Topological Insight into Viral Evolution
Abstract: Biological sequences are often represented using the frequencies of short subsequences, or k-mers. While these representations are computationally convenient, they provide limited information about how k-mers are organized and related within a sequence. In this talk, I will present a topological framework for characterizing the multiscale organization of k-mers in genomic sequences. By combining persistent homology with spectral information from the persistent Laplacian, the resulting representation captures both topological and geometric features that are not reflected by k-mer frequencies alone. I will demonstrate how these descriptors can be used to compare and classify viral genomic sequences and to construct quantitative measures of genetic similarity. I will conclude by discussing how this framework can be extended to develop more general structure-aware representations for biological sequence data.