Mots-Clés

metabolism metabolic modelling metabolic networks R RShiny

Description

Context

Metabolism is a set of crucial biological processes which ensures that cells have the energy and components required to survive, function, and grow. It consists of the set of chemical reactions which break down molecules into smaller compounds known as metabolites. Being able to experimentally observe and measure the presence, quantities and variations of metabolites is of major importance in human health, as many diseases are linked to parts of this system being disrupted. However, there are limitations in our current ability to observe the entire set of metabolites present in an experimental sample. Computational approaches, such as metabolic modelling, can help bridge these gaps by simulating metabolism.

The known metabolism of an organism can be modelled in the form of a genome-scale metabolic network, containing interconnected sets of metabolites, reactions, and metabolic genes [1]. We can then create a disruption in this metabolic network, to simulate any sort of metabolic perturbation. SAMBA [2], a constraint-based modelling approach developed within the team, can use this disrupted network to simulate the impact on all imported and exported metabolites (metabolic profile) expected to occur in that metabolic state.

The results simulated by SAMBA consist of z-scores for each exchange (exported/imported) metabolite in the network, which is around 1500 metabolites for the Human1 metabolic network for example [1]. These z-scores reflect how much more or less the metabolite was exported after a disruption occurs in the network. The goal is to help users to analyse the results from SAMBA easily and efficiently, by creating an accessible tool for both biologists and bioinformaticians. A basic RShiny app was developed for this purpose, but the current version does not handle variations in data types, and is limited to running analyses according to the initial simulated conditions SAMBA was developed for. A flexible and more informative RShiny application would be of great use to the community in managing and analysing the results from SAMBA, and would help to bridge the gap between experimental metabolic profiles and simulated metabolic profiles.

Internship objective: The main objective of the internship is to build upon the existing RShiny application that was created alongside SAMBA during its development. This RShiny app should be able to handle the new types of data and features implemented in SAMBA since its initial release, as well as generate new plots and visualisations of the results. During this project, you will: • Learn the basics of metabolic modelling and constraint-based modelling for flux simulation, to understand SAMBA’s inputs and outputs. • Build upon and develop an RShiny application which takes user-uploaded files as input, and runs analyses and plots. • Learn R libraries such as ggplot2 and ggraph, and the tidyverse collection. 1• Understand the parallels and differences of simulated metabolic profiles and experimental metabolic profiles, and how the results of the simulations can be applied to experimental questions.

Skills: We are looking for a Master’s 1 (first year) student in bioinformatics, computational biology or equivalent, with the following skills: • Experience with R • Knowledge of biology/systems biology • Experience with git • Scientific English The following skills would be a bonus: • Knowledge of metabolic networks and/or metabolomics • Experience with RShiny

Work environment: The internship will take place in our bioinformatics team (within the MeX team), supervised by Juliette Cooke (INRAE post-doc) and Nathalie Poupin (INRAE Researcher) in the INRAE Toxalim UMR1331 laboratory in Toulouse, for 2-3 months in 2025.

Contact information: Please send your CV and cover letter to juliette.cooke@inrae.fr and nathalie.poupin@inrae.fr

Useful links:

• Current RShiny app: https://samba.sk8.inrae.fr/

References:

[1] Jonathan L. Robinson et al. “An Atlas of Human Metabolism”. In: Science Signaling 13.624 (Mar. 2020), eaaz1482. issn: 1937-9145. doi: 10.1126/scisignal.aaz1482.

[2] Juliette Cooke et al. “Genome Scale Metabolic Network Modelling for Metabolic Profile Predictions”. In: PLOS Computational Biology 20.2 (Feb. 2024), e1011381. issn: 1553-7358. doi: 10.1371/journal.pcbi.1011381.