Section: New Results

Characterization of Molecular Biodiversity

In 2019 Pleiade developed new methods for characterizing molecular biodiversity (see [4], [5] for applications). This point itself has been developed with two approaches in 2019, each with the beginning of a PhD.

  • Building OTUs from a pairwise distance matrix typically is an unsupervised clustering issue. In this new development, the PhD student (Pleiade ) tests whether SBM (Stochastic Block Model) approach yields relevant results for a global characterization of biodiversity beyond a summary by a scalar index. This is done in collaboration with MIAT INRAE research unit in Toulouse and EPC HiePACS. It represents a connection between metabarcoding and statistical modeling, a topic which deserves investigation and is expanding. (Figure 5 from [8])

  • A major goal of Pleiade is to develop a geometric view on biodiversity. The tool selected up to now is to associate a point cloud to a dataset (pairwise distances between sequences) and study its shape. In 2019, Pleiade has been associated in a collaboration with HiePACS to begin a new topic: comparison between point clouds, each cloud being associated to a data set. Indeed, the development of metabarcoding leads to the new issue of comparison between OTUs built from different dataset. This approach is part of the issues raised in a PhD supervized by HiePACS, in collaboration with Pleiade .

Figure 5. Graphical representation of a coupled HMM with three hidden chains (from [8])