De novoGene Signature Identification from Single-Cell RNA-Seq with Hierarchical Poisson Factorization

Abstract

Common approaches to gene signature discovery in single cell RNA-sequencing (scRNA-seq) depend upon predefined structures like clusters or pseudo-temporal order, require prior normalization, or do not account for the sparsity of single cell data. We present single cell Hierarchical Poisson Factorization (scHPF), a Bayesian factorization method that adapts Hierarchical Poisson Factorization [1] for de novo discovery of both continuous and discrete expression patterns from scRNA-seq. scHPF does not require prior normalization and captures statistical properties of single cell data better than other methods in benchmark datasets. Applied to scRNA-seq of the core and margin of a high-grade glioma, scHPF uncovers marked differences in the abundance of glioma subpopulations across tumor regions and subtle, regionally-associated expression biases within glioma subpopulations. scHFP revealed an expression signature that was spatially biased towards the glioma-infiltrated margins and associated with inferior survival in glioblastoma.

Identifying gene expression programs of cell-type identity and cellular activity with single-cell RNA-Seq

“…For example, cNMF does not specifically address the count nature of gene expression. Recently developed statistical frameworks that address these aspects of scRNA-Seq data such as Hierarchical Poisson Factorization (Levitin et al, 2018) may therefore increase the accuracy of GEP inference. In addition, NMF often yields low but non-zero usages for many GEPs even though we expect most cells to express a small number of identity and activity GEPs.…”

The Whole Body as the System in Systems Immunology

“…Unsupervised clustering of scRNA-seq data from bone marrow, lung, and LLN from two organ tissue donors revealed three unique CD4 + T cell activation states distinguished by differential expression of IL2, TNF, and IL4R and two major functional states for CD8 + T cells characterized by differential expression of genes associated with pro-inflammatory cytokines and chemokines and cytotoxic mediators ( Szabo et al., 2019a ). Applying single-cell Hierarchical Poisson Factorization (scHPF) ( Levitin et al., 2019 ) revealed activation and functional “modules” that were highly conserved across tissues and donors, including a proliferation module, an IFN response module, and two unique cytotoxicity modules. Moreover, projection of blood T cells profiles onto UMAP embeddings of T cells from tissue donors revealed that tissue T cells upregulated genes associated with cell structure, extracellular matrix, adhesion, and tissue residency, suggesting that T cells adopt structural changes that facilitate interactions with tissue matrix ( Szabo et al., 2019a ).…”

A single-cell reference map for human blood and tissue T cell activation reveals functional states in health and disease

“…We began by clustering the single-cell profiles within each sample using a pipeline that we reported previously 65,66 . Briefly, we identified highly variable genes that are likely markers of specific subpopulations by normalizing the molecular counts for each cell to sum to one, ordering all genes by their normalized expression values, and computing a drop-out score ds g for each gene g defined as:…”

“…We applied Single-cell Hierarchical Poisson Factorization (scHPF), a method that we recently reported for de novo discovery of gene expression signatures in scRNA-seq data, to the merged 2 7 activated and resting cells for each tissue and donor 66 . Given a molecular count matrix, scHPF identifies a small number of latent factors that explain both continuous and discrete expression patterns across cells.…”