Novel machine learning challenges for modeling cell developmental dynamics

Abstract In this talk I will introduce trajectory inference methods, a novel class of unsupervised computational techniques to model dynamic processes. These techniques are generally applicable, but a particular case study that will be highlighted in the talk is the application to high-throughput, single-cell data.  Recent advances in single-cell analysis, such as 20+color flow cytometry, mass cytometry and single-cell transcriptomics allow scientists to measure an increasing number of parameters per cell, generating large and high-dimensional datasets. Trajectory inference methods offer a complementary view to existing techniques such as clustering and dimensionality reduction methods. An overview of current state-of-the-art methods as well as novel developments in the field will be presented.

About the speaker Yvan Saeys is associate professor of Machine Learning and Systems Immunology at VIB and Ghent University.  He is developing state-of-the-art data mining and machine learning methods for biological and medical applications, and is an expert in computational models to analyse high-throughput single-cell data.  The methods he develops have been shown to outperform competing techniques, including computational techniques for regulatory network inference (best performing team at the DREAM5 challenge) and biomarker discovery from high-throughput, single cell data (best performing team at the FlowCAP-IV challenge).  Yvan Saeys has published >180 papers in top ranking journals and conferences, ranging from methodological development in machine learning and bioinformatics to applications in cancer, immunology and medicine.