Functional inference integrated in the FROGS suite

Moussa Samb (1) , Maria Bernard (2) , Géraldine Pascal

1 INRAE - Institut National de Recherche pour l’Agriculture, l’Alimentation et l’Environnement
2 GABI - Génétique Animale et Biologie Intégrative

The high-throughput sequencing of biomarkers has opened new horizons in the study of microbial communities. To help biologist in their studies, several years ago, we developed FROGS [1] is a metabarcoding analysis pipeline. It gives, among other information, the abundance table, the taxonomic affiliation of operational taxonomic units (OTUs) and statistics data. In addition to command line mode, it can be used as a Galaxy workflow, focused on user-friendliness, so it does not require bioinformatics or command lines skills. To go further in their analyses, biologists generally need metabolic data associated with the microbial composition of the environment they are studying. There are currently several solutions: (i) analysis by RNASeq, (ii) analysis by metagenomics sequencing (iii) analysis by metabolomics and (iv) analysis by functional inference. Despite the ever-increasing accessibility of metagenomics sequencing, functional inference from data obtained from amplicons remains very useful. Indeed, this strategy is important for samples with high host contamination, low biomass and when metagenomic sequencing is not economically feasible. In any case, all produced results by these kind of methods should primarily be used for hypothesis generation. The principle of functional inference is to infer the metabolic pathways of organisms based only on their taxonomic affiliation. Several tools can do this, MACADAMExplore [2], Tax4Fun [3], PAPRICA [4], Piphillin [5] and PICRUSt2 [6] for example. PICRUSt2 bases on marker gene from sequencing profiles. First, it relies on an algorithm that insert marker sequence into an existing phylogenetic tree thank to short-read placement tools. After, it infers gene family of OTUs. Then, it determines gene family abundance per sample. Finally, it infers pathway abundances to predict sample pathway abundances. It enables functional predictions from 16S or 18S or ITS amplicon profiling. Thus, to allow the FROGS community to deduce the potential metabolic functions of the targeted environment, we have developed a series of applications in python using PICRUSt2 in particular. The Galaxy interface of these applications also allows non-expert users to use easily these new features.

Domaines: Bio-informatique [q-bio.QM]

Moussa Samb, Maria Bernard, Géraldine Pascal. Functional inference integrated in the FROGS suite. JOBIM2020, Jun 2020, Montpellier, France. , ⟨10.1038/s41587-⟩. ⟨hal-03176828⟩