RT Journal Article
SR Electronic
T1 BGP: Branched Gaussian processes for identifying gene-specific branching dynamics in single cell data
JF bioRxiv
FD Cold Spring Harbor Laboratory
SP 166868
DO 10.1101/166868
A1 Alexis Boukouvalas
A1 James Hensman
A1 Magnus Rattray
YR 2017
UL http://biorxiv.org/content/early/2017/08/01/166868.abstract
AB High-throughput single-cell gene expression experiments can be used to uncover branching dynamics in cell populations undergoing differentiation through use of pseudotime methods. We develop the branching Gaussian process (BGP), a non-parametric model that is able to identify branching dynamics for individual genes and provides an estimate of branching times for each gene with an associated credible region. We demonstrate the effectiveness of our method on both synthetic data and a published single-cell gene expression hematopoiesis study. The method requires prior information about pseudotime and global cellular branching for each cell but the probabilistic nature of the method means that it is robust to errors in these global branch labels and can be used to discover early branching genes which diverge before the inferred global cell branching. The code is open-source and available at https://github.com/ManchesterBioinference/BranchedGP.