Stan Webinar- hosted by Immunobiology and Infection SMB Subgroup

The Immunobiology and Infection subgroup is pleased to announce a 2-part methods webinar this coming August for attendees interested in learning to use Stan, taught by Dr. Christiaan van Dorp (Los Alamos National Laboratory).

Stan is a powerful platform for statistical modeling and high-performance statistical computation. It is used primarily for Bayesian statistical modeling, data analysis, parameter estimation including for ODE models, and to make predictions, in the social, biological, and physical sciences. More details on Stan below.

DATES & TIMES:
August 3 12:30-2:30 EDT
August 5 12:30-2:30 EDT

HOW TO REGISTER:
Registration is free and SMB/Subgroup membership is not required.

Please complete the form available at: LINK

The registration deadline is August 2nd. If you’d like to register after that, please contact Jessica Conway (jmconway@psu.edu<mailto:jmconway@psu.edu>).

Following registration, you’ll get an email with a link where course materials will be posted. Installation instructions and full course materials including Jupyter notebooks will also be posted on that site nearer the webinar date.

As we near the date you will also me emailed the webinar zoom link.

For further information please contact Jessica Conway (Penn State), jmconway@psu.edu<mailto:jmconway@psu.edu>.

—-
WHY STAN?

Stan is a powerful platform for Bayesian statistical modeling. Bayesian statistics has certain benefits over frequentist approaches, such as the possibility to incorporate prior knowledge in the model, and access to the full posterior density of the parameters. However, in most use cases, Bayesian inference requires Monte Carlo methods that are computationally intensive and time consuming. One aspect that makes Stan so appealing is a very effective implementation of the highly efficient Hamiltonian Monte Carlo algorithm which to a large extent automatically tunes algorithmic parameters.

The second appeal of Stan is that it provides a highly expressive and flexible programming language with built-in support for many statistical distributions, and auxiliary functions such as ODE integrators.

The versatility of the Stan language allows for the implementation of highly complex models that are common in immunobiology. Often, we encounter multi-dimensional time-series data with repeated experiments and multiple covariates. However, for modelers that are new to Stan, implementing such a model can be very challenging. The webinar starts with the basics, and will then gradually move to more complex models with examples from e.g. viral dynamics and epidemiology.

Dr. van Dorp will provide Jupyter notebooks to work through the examples discussed in the webinar.

—
ABOUT DR. VAN DORP

Christiaan van Dorp has a PhD in theoretical biology from Utrecht University and a MSc degree in mathematics from the University of Amsterdam. Currently, he is a postdoc at Los Alamos National Laboratory in the group of Alan Perelson. Chris is working on topics in epidemiology and viral dynamics and is using Stan in many of his projects.