19-23 March 2018

Dimensions reduction and visualization techniques for understanding genomic data.

This course is divided into two parts.

First, a detailed overview of the classical exploratory methods conceived for multivariate data: Principal Components Analysis, Correspondence Analysis, and Multiple Correspondence Analysis. From a unified theoretical framework, we will see how these methods are linked, as well as their specificities in terms of interpretation, due to the nature of the data they are dealing with. From a practical point of view, we will see how they can be applied to genomic data, and how they can be used to obtain meaningful information. We will see notably, how we can add supplementary information to get a better understanding of the data.

Second, an overview of methods that handle multivariate data, when variables are structured according to groups: generalised canonical analysis, and Multiple Factor Analysis. These methods are really useful when different points of view on the same set of individuals have to be compared. It is the case for instance, when one has at his disposal gene expressions on the one hand, and chemical measures on the other hand. 

The methods will be presented from a geometrical point of view. The concepts of quality of representation, active versus illustrative variables, automatic description of the dimensions provided by the analyses will be discussed.

Each day will include an introductory lecture with class discussion of key concepts. The remainder of each day will consist of practical hands-on sessions. These sessions will involve a combination of both mirroring exercises with the instructor to demonstrate a skill as well as applying these skills on your own to complete individual exercises. After and during each exercise, interpretation of results will be discussed as a group. Computing will be done using a combination of tools installed on the attendees laptop computer and web resources accessed via web browser.

Researchers who would like to investigate multivariate and heterogenous data from an exploratory point of view. Researchers who would like to invest in methods capable of handling multi-block data, in the sense that data are structured into groups of variables.

Botanisches Museum, Königin-Luise-Straße 6-8, Berlin

Monday, 9.30-17.30

Monday – Classes from 09:30 to 17:30

Reducing dimensionality, the general framework.

An introduction to the notion of multivariate data, to the concept of inertia, to the analysis of multivariate data. From the analysis of the rows to the analysis of the columns: the transition formulae. What is an illustrative variable and how to use it?

Combining the best of two worlds.

Principal Component Analysis, the « usual » way and the « French » way. My first PCA on gene expression data, with R: practical overview of multivariate packages.

Please click HERE to get all the information about our packages.

Registration deadline: 20th February 2018.