SI2-SSE: Algorithms and Tools for Data-Driven Executable Biology

University Of Washington, Seattle WA

Investigators

Abstract

This project seeks to understand the signaling mechanisms that control cellular activities such as cell division, cell growth, and cell differentiation. Errors in cellular signaling cause diseases such as cancer, autoimmunity, and diabetes. Accurate models of cellular signalling are thus necessary for rational drug design and other applications central to national health. This project focuses on inferring models from experimental data. Specifically, it is interested in models of protein signalling because proteins control and mediate the vast majority of biological processes in a living cell. The project follows of the approach of executable biology: models of cell signalling are computer programs, which allows executing the models on the computer and comparing the model behavior against the behavior of the living cell observed in the lab setting. Most importantly for this project, viewing models as programs will allow the team to harness the recent advances in automatic synthesis of computer programs for synthesis of models from experimental measurements of cells. The goal of the project is to provide biologists with a tool that synthesizes a variety of executable models from varied types of experimental data. To facilitate synthesis of mechanistic models from experimental data, the project will develop a family of modeling languages that will capture complex behaviors of biological systems, such as time and concurrency. The languages will be instances of the more general Boolean-Networks language. The team will investigate how to adjust the modeling abstraction based on the nature of available experimental data; the abstractions will be instantiated as suitably chosen languages from their language family. The modeling framework will be built by leveraging techniques from programming languages and formal methods such as meta-programming and constraint solving.