Supporting Environmental Decision-Making: Modeling Complex and Noisy Biology

Everyone wants assurance that the chemicals we bring into our homes, such as those in cosmetics, cleaning products, pesticides, and food ingredients, are safe. However, achieving these assurances is daunting; it is estimated that people are exposed to 30,000–40,000 different manmade chemicals in their lifetimes, and only a fraction of these have ever been thoroughly tested in traditional modeling systems (e.g., rats, mice, rabbits). The “too many chemicals” problem will not be solved through traditional animal testing because of cost and animal welfare concerns. The emerging alternative is to use a combination of experiments with cell systems or model organisms (e.g., zebrafish embryos) and mathematical/computational models. This session discusses the promise of mathematical modeling with three case studies. Through examples, speakers will address the following key challenges: models must be able to integrate data from simple levels of biological organization and predict effects on whole animals; methods must be scalable to be able to make predictions on tens to hundreds of thousands of chemicals; and methods must be robust in the face of noise in both the simple data on which they are built and the complex phenotypes against which they are validated.