La biologie est de plus en plus quantitative. Cette évolution passe par des jeux de donnée toujours plus grand à analyser et les méthodes issue des 'sciences des données' au sens large prennent une place de plus en plus grande dans la biologie. Dans cet EA nous verront les méthodes les plus importantes des sciences des données et les appliqueront dans le contexte du phénotypage par microscopie haut débit/haut contenu.

Déroulement du cours: 6 séance cours/TP durant lesquelles sera ré-analysé un jeu de donnée issu d'une publication de recherche nous permettant de voir les méthodes correspondant à toutes les étapes d'un workflow de phénotypage haut contenu: traitement et analyse d'image, calcul de caractéristiques quantitatives, analyses statistiques, apprentissage statistique et 'deep learning' et lien à la biologie computationel au sens large via ontologies et base de donnée partagé en ligne. Suivront 3 séances projets pour aller plus loin sur une problématique particulière, sur ce jeu de donnée ou un autre. Évaluation: oral devant jury sur le projet.

Pré-requis: Formellement aucun, pour permettre à tout élève intéressé de suivre l'EA, même en dehors des PA bio/bioinfo. Un réel intérêt pour la biologie et un minimum d'aisance en programmation en python sont néanmoins nécessaire.

Langue du cours : Anglais
Credits ECTS : 4

Data sciences of biological imaging: Image-based quantitative phenotyping

Biology is increasingly quantitative. In particular, ever-bigger datasets are acquired and sophisticated methods from the wider data sciences are now a part of the quantitative biology toolbox. In this EA we will see a wide range of data science techniques in the context of image based quantitative phenotyping with high content/high throughput microscopy.

Plan: 6 session of courses/practicals during which a published datasets from an academic paper will be reanalysed, allowing us to cover all step of such an analysis pipeline: image processing, image analysis, quantitative features computation, statistical analysis, machine learning and 'deep learning', and links with computational biology at large with ontologies and online databases. They will be followed by 3 more open project sessions to go further into one particular topic or another dataset. Evaluation will be an oral on the project.

Requirements: Formally none, to allow any interested student to potentially enrol. In practice, a real interest for biology in general, and a certain familiarity with the Python programming language will be needed.

Langue du cours : English
Credits ECTS : 4




Data sciences of biological imaging: Image-based quantitative phenotyping

Biology is increasingly quantitative. In particular, ever-bigger datasets are acquired and sophisticated methods from the wider data sciences are now a part of the quantitative biology toolbox. In this EA we will see a wide range of data science techniques in the context of image based quantitative phenotyping with high content/high throughput microscopy.

Plan: 6 session of courses/practicals during which a published datasets from an academic paper will be reanalysed, allowing us to cover all step of such an analysis pipeline: image processing, image analysis, quantitative features computation, statistical analysis, machine learning and 'deep learning', and links with computational biology at large with ontologies and online databases. They will be followed by 3 more open project sessions to go further into one particular topic or another dataset. Evaluation will be an oral on the project.

Requirements: Formally none, to allow any interested student to potentially enrol. In practice, a real interest for biology in general, and a certain familiarity with the Python programming language will be needed.