In high-throughput biological data, dynamic correlation, i.e. changing correlation patterns under different biological conditions, can reveal important regulatory mechanisms. Current methods seek underlying conditions of dynamic correlation by using certain genes as surrogate signals. Here we describe a new method that directly identifies strong underlying signals that regulate the dynamic correlation of many pairs of genes, named LDCA: Latent Dynamic Correlation Analysis. We validate the performance of the method with extensive simulations. In real data analysis, the method reveals biologically plausible latent signals that are not found by existing methods.

Tianwei Yu is an Associate Professor at the Department of Biostatistics and Bioinformatics of Emory University. He received his B.S. degree from the Department of Biological Sciences of Tsinghua University, and his Ph.D. degree in Statistics from the University of California, Los Angeles (UCLA). He started working as a faculty member at Emory University since 2006. His current research interests include high-dimensional data analysis, biomarker discovery, biological networks, and data pre-processing in metabolomics.