Recent machine learning approaches for single-cell RNA-seq data analysis

Abstract

DNA sequencing has become an extremely popular assay with researchers claiming that in the distant future, the DNA sequencing impact will be equal to the microscope impact. Single-cell RNA-seq (scRNA-seq) is an emerging DNA-sequencing technology with promising capabilities, but with major computational challenges due to the large-scaled generated data. Given the fact that sequencing costs are constantly decreasing, the volume and complexity of the data generated by these technologies will be constantly increasing. Toward this direction, major computational challenges are posed at the cell level, in particular, when focusing on the ultra-high dimensionality aspect of the scRNA-seq data. The main challenges are related to three pillars of machine learning (ML) analysis, classification, clustering, and visualization methods. Although there has been remarkable progress in ML methods for single-cell RNA-seq data analysis, numerous questions are still unresolved. This review records the state-of-the-art classification, clustering, and visualization methods tailored for single-cell transcriptomics data. © Springer-Verlag GmbH Germany, part of Springer Nature 2020.