Section: New Results

Scaling Metabarcoding Programs

Metabarcoding is a series of technical procedures to build molecular based inventories from large datasets of amplicons. We derived new methods and tools to scale metabarcoding programs in collaboration with EPC HiePACS. This has been realized through following participation in research projects:

• Contribution to ADT Gordon project in Inria BSO. The objective of this project (partners: Tadaam (coordinator), STORM, HiePACS, Pleiade ) is to integrate SVD as an available tool in Chameleon, starPU and new Madeleine. The contribution of Pleiade is to bring metabarcoding as a use case, and random projection ([6]) as a method for scaling Multidimensional Scaling (which requires an SVD) in collaboration with HiePACS with a template implemented in Diodon. In 2019, Pleiade has bought to the project a series of 55 matrices with size about ${10}^5$ rows/column which, assembled, yield a full pairwise distance matrix between one million of sequences. The objective is to reach the million in 2020.

• Contribution to Region Nouvelle Aquitaine project “HPC Scalable Ecosystem.” This project is chaired by HiePACS. In collaboration with this EPC, Pleiade is involved in developing a new approach for comparing OTUs built from different datasets.