scClassify: hierarchical classification of cells

Abstract

17Cell type identification is a key computational challenge in single-cell RNA-sequencing 18 (scRNA-seq) data. To capitalize on the large collections of well-annotated scRNA-seq datasets, 19 we present scClassify, a hierarchical classification framework based on ensemble learning. 20 scClassify can identify cells from published scRNA-seq datasets more accurately and more 21 finely than in the original publications. We also estimate the cell number needed for accurate 22 classification anywhere in a cell type hierarchy. 23 24Single cell RNA-sequencing (scRNA-seq) datasets have become larger and more diverse, driving the 25 need for classification methods that can provide more refined discrimination between cell types. 26Many major cell types can be divided into subtypes in a hierarchical fashion, forming what we call a 27 'cell type hierarchy' 1,2 . Approaches to scRNA-seq studies that account for cell type hierarchies in 28 experimental design and cell type identification will permit more nuanced interpretations of the 29 resulting data. 3 30 31The most common approach to identifying cell types within scRNA-seq data is unsupervised 32 clustering 4-6 followed by manual annotation using known marker genes. However, the number of 33

dyngen: a multi-modal simulator for spearheading new single-cell omics analyses

“… For N = 1, P ( S 1 ) is equal to the Hill equation, where k i represents the concentration of H i at half-occupation and n i represents the Hill coefficient. Typically, n i is between [1,10] The Hill equation can be simplified by letting . Since P ( S 0 ) = 1 − P ( S 1 ), the activation function is formulated and simplified as follows.…”

“…Other generators of scRNA-seq data (e.g. splatter [1], powsimR [4], PROSSTT [5] and SymSim [6]) have already been used extensively to explore the strengths and weaknesses of computational tools, both by method developers [7, 8, 9, 10] and independent benchmarkers [11, 12, 13]. However, a limitation of these existing simulators is that they would require significant methodological alterations to add additional modalities or experimental conditions (Table S1).…”

Northstar enables automatic classification of known and novel cell types from tumor samples

“…writing emails to the authors). Cell type annotations are difficult to find and often of mediocre quality, presenting a challenge in particular to algorithms such as scClassify that rely on entire cell type hierarchies 17 . Although northstar is less affected because it does not require a hierarchical ontology, we aspire to change this trend by providing a website with averages and subsamples for several atlases that can be accessed both by humans and programmatically: https ://north stara tlas.githu b.io/atlas _landm arks.…”

Improving replicability in single-cell RNA-Seq cell type discovery with Dune

“…In practice, these methods define marker genes for known cell types and build classifiers to assign new cells to these cell types. In particular, Garnett [29] allows a hierarchical clustering structure, but one that needs to be predefined, and scClassify [30] uses the HOPACH [31] algorithm to establish a hierarchy in the training dataset. Most of these algorithms can also identify new cell types not present in the reference.…”