Big data analytics in single‐cell transcriptomics: Five grand opportunities

Abstract

Single‐cell omics technologies provide biologists with a new dimension for systematically dissecting the underlying complexities within biological systems. These powerful technologies have triggered a wave of rapid development and deployment of new computational tools capable of teasing out critical insights by analysis of large volumes of omics data at single‐cell resolution. Some of the key advancements include identifying molecular signatures imparting cellular identities, their evolutionary relationships, identifying novel and rare cell‐types, and establishing a direct link between cellular genotypes and phenotypes. With the sharp increase in the throughput of single‐cell platforms, the demand for efficient computational algorithms has become prominent. As such, devising novel computational strategies is critical to ensure optimal use of this wealth of molecular data for gaining newer insights into cellular biology. Here we discuss some of the grand opportunities of computational breakthroughs which would accelerate single‐cell research. These are: predicting cellular identity, single‐cell guided in silico drug screening for precision medicine, transfer learning methods to handle sparsity and heterogeneity of expression data, establishing genotype–phenotype relationships at single‐cell resolution, and developing computational platforms for handling big data. This article is categorized under: Algorithmic Development > Biological Data Mining Fundamental Concepts of Data and Knowledge > Big Data Mining Technologies > Machine Learning

scSPARKL: Apache Spark based parallel analytical framework for the downstream analysis of scRNA-seq data

“…With the advancement of the cell profiling techniques and rapid cost drops, scientists are now able to sequence millions of cells in parallel. 10x chromium profiling, for instance, provides the capability of profiling more than 80000 cells in one experiment parallelly (Bhattacharya et al, 2021). For each cell measurements of more than 20,000 genes are taken, generating a Cell x gene matrix of an enormous size.…”

Pseudo-grading of tumor subpopulations from single-cell transcriptomic data using Phenotype Algebra

“…Thus, the phenotypic characterization of malignant cell subpopulations (subclones) is critical to understanding the underlying mechanisms of resistive behavior. The widespread adoption of single-cell RNA-sequencing (scRNA-seq) has enabled the profiling of individual cells, thereby obtaining a high-resolution snapshot of their unique molecular landscapes 3,4 . A precise understanding of cell-to-cell functional variability captured by scRNA-seq profiles is crucial in this context.…”