Construction of Correlation Networks with Explicit Time-Slices Using Time-Lagged, Variable Interval Standard and Partial Correlation Coefficients

The construction of gene regulatory models from microarray time-series data has received much attention. Here we propose a method that extends standard correlation networks to incorporate explicit timeslices. The method is applied to a time-series dataset of a study on gene expression in the developmental phase of zebrafish. Results show that the method is able to distinguish real relations between genes from the data. These relations are explicitly placed in time, allowing for a better understanding of gene regulation. The method and data normalisation procedure have been implemented using the R statistical language and are available from http://zebrafish.liacs.nl/supplements.html.