Abstract:
|
Community detection, which involves identification of the number of clusters in a network and the membership in each, is a challenging problem, especially in applications like gene co-expression when the information about the network is uncertain. We propose a global community detection method that combines information across a series of networks, longitudinally, to strengthen the inference for each period. Our method is derived from evolutionary spectral clustering and degree correction methods. Data driven solutions to the problem of tuning parameter selection are provided. Recently obtained data from rhesus monkey brains provide a dense temporal sampling of prenatal and postnatal periods, with samples from finely partitioned brain regions. Of interest is how gene communities develop over space and time; however, once the data are divided into spatial and temporal periods, sample sizes are very small, making inference quite challenging. In both simulations and the brain data we find that our approach performs much better than competing methods designed with a low signal-to-noise ratio. Our methods are designed for application to single cell RNAseq data.
|