Modeling biological data

This course is a natural continuation of the " Quantifying Hazards " course (HAV424B) presented in S4. It should provide the concepts for constructing experimental protocols that answer biological questions, and for associating appropriate models for analyzing variability. The first part will be devoted to the construction of experimental protocols that can answer a multitude of questions in the life sciences, i.e. taking into account the inevitable dependence of statistical individuals, such as pairing and the spatial or temporal structure of populations. This part of the course will also introduce the notion of fluctuation, replication and pseudo-replication, which will be taken into account in the models built in the second part of the course. The second part of the course will show the link between the experimental protocol and the modeling of the variability of a quantitative response variable, through the construction of models including several qualitative or quantitative variables. Particular attention will be paid to the conditions of application of these methods, to type I and type II errors, to methods for estimating the parameters of the models constructed (including likelihood) and to the interpretation of the estimated parameters. Each notion will be illustrated by the analysis of real biological data from a variety of themes, helping students to discover not only modern and current biological questions, but also the tools developed to answer them. Practical work in R will enable students to independently carry out analyses on published biological cases.

- HAV312B, HAV313B and HAV424B

- Mobilize mathematical tools to build statistical models of variability and respond to life science issues

- adjust the protocol to the question asked

- fit a model to experimental data

- validate model predictions and assess model limitations