Dimensionality reduction and statistical modeling of scGET-seq data

Abstract

Single cell multiomics approaches are innovative techniques with the ability to profile orthogonal features in the same single cell, giving the opportunity to dig more deeply into the stochastic nature of individual cells. We recently developed scGET-seq, a technique that exploits a Hybrid Transposase (tnH) along with the canonical enzyme (tn5), which is able to profile altogether closed and open chromatin in a single experiment. This technique adds an important feature to the classic scATAC-seq assays. In fact, the lack of a closed chromatin signal in scATAC: (i) restricts sampling of DNA sequence to a very small portion of the chromosomal landscapes, substantially reducing the ability to investigate copy number alteration and sequence variations, and (ii) hampers the opportunity to identify regions of closed chromatin, that cannot be distinguished between non-sampled open regions and truly closed. scGET-seq overcomes these issues in the context of single cells. In this work, we describe the latest advances in the statistical analysis and modeling of scGET-seq data, touching several aspects of the computational framework: from dimensionality reduction, to statistical modeling, and trajectory analysis.