EnClaSC: a novel ensemble approach for accurate and robust cell-type classification of single-cell transcriptomes

Abstract

Background In recent years, the rapid development of single-cell RNA-sequencing (scRNA-seq) techniques enables the quantitative characterization of cell types at a single-cell resolution. With the explosive growth of the number of cells profiled in individual scRNA-seq experiments, there is a demand for novel computational methods for classifying newly-generated scRNA-seq data onto annotated labels. Although several methods have recently been proposed for the cell-type classification of single-cell transcriptomic data, such limitations as inadequate accuracy, inferior robustness, and low stability greatly limit their wide applications. Results We propose a novel ensemble approach, named EnClaSC, for accurate and robust cell-type classification of single-cell transcriptomic data. Through comprehensive validation experiments, we demonstrate that EnClaSC can not only be applied to the self-projection within a specific dataset and the cell-type classification across different datasets, but also scale up well to various data dimensionality and different data sparsity. We further illustrate the ability of EnClaSC to effectively make cross-species classification, which may shed light on the studies in correlation of different species. EnClaSC is freely available at https://github.com/xy-chen16/EnClaSC. Conclusions EnClaSC enables highly accurate and robust cell-type classification of single-cell transcriptomic data via an ensemble learning method. We expect to see wide applications of our method to not only transcriptome studies, but also the classification of more general data.

Cofea: correlation-based feature selection for single-cell chromatin accessibility data

“…First, considering that a substantial portion of scCAS data analytic tools is implemented in the R, we will develop an R version of Cofea, to offer researchers a richer pool of alternatives. Second, such a correlation-based framework can also be extended to scRNA-seq data or single-cell multi-omics data, thereby enhancing the performance of downstream analysis, such as cell type annotation [ 46 ]. It is interesting to consider how to incorporate the gene–gene or gene–peak intrinsic relationship with the feature selection step.…”

Cell-type annotation with accurate unseen cell-type identification using multiple references

“…Noise from the reference data and incorrectly annotated cell types may lead to inaccurate annotations on the query data, and the selection of input features of the classification model can also impact the annotation performance of different methods [20,21]. This issue can be partially addressed by integrating multiple well-annotated reference datasets and multiple gene selection methods [22][23][24][25], but an appropriate integration strategy is needed. Previous methods often integrate multiple well-labeled datasets to create a comprehensive reference atlas, which is then used to annotate the cell types in new data.…”

CellAnn: a comprehensive, super-fast, and user-friendly single-cell annotation web server

“…Some tools are only available in specific computational languages such as Python or R. Third, an ideal method should run fast even with large reference or query datasets, and it should not require a large computer memory. Some advanced methods, such as those employing deep learning approaches, have been developed and have a good performance [scBERT ( Yang et al 2022 ), scDeepSort ( Shao et al 2021 ), ACTINN ( Ma and Pellegrini 2020 ), sigGCN ( Wang et al 2021 ), scIAE ( Yin et al 2022 ), scNym ( Kimmel and Kelley, 2021 ), SuperCT ( Xie et al 2019 ), and EnClaSC ( Chen et al 2020 )]. However, these methods are often slow and require large memories and computing resources, making them not suitable for an online tool.…”