Step 3: Using Time-Series Data

Abstract

Time-series data gives information about the values of genes at a series of consecutive time points. This temporal information can be exploited to infer directionality of edges, or help to infer causal relations between genes. However, adding temporal information also creates a more complex dataset. It adds interdependencies between experiments (time-points) that don’t exist in steady-state data, so more care has to be taken in analysis. Three types of algorithms will be presented in this section: mutual information, ordinary differential equations with l1 regularization, and dynamic Bayesian Networks. Each of these approaches makes different assumptions about the data.