1 min read

Kernel methods in systems biology

Jérôme Mariette : Kernel methods in systems biology Plateforme Genotoul Bioinfo, Unité de Mathématique et Informatique Appliquées de Toulouse, INRAE

The development of high-throughput sequencing technologies has lead to produce high dimensional heterogeneous datasets at different living scales. To process such data, integrative methods have been shown to be relevant, but still remain challenging. Here, we propose contributions useful to simultaneously explore heterogeneous multi-omics datasets and to select relevant features in a dataset. To tackle these problems, kernels and kernel methods represent a natural framework because they allow to handle the own nature of each datasets while permitting their combination. In a first part, I will present a multiple kernel framework that allows to integrate multiple datasets of various types into a single exploratory analysis. In a second part, I will introduce feature selection methods that are adapted to the kernel framework and go beyond the well established work in supervised learning by addressing the more difficult tasks of unsupervised learning and kernel output learning. The proposed methods efficiency is highlighted in the domain of microbial ecology : eight TARA oceans,datasets are integrated and analysed.

References

package R mixKernel site mixKernel
Mariette, J. and Villa-Vialaneix, N. (2018). Unsupervised multiple kernel learning for heterogeneous data integration. Bioinformatics, 34(6), 1009-1015