About this tutorial

In this tutorial, we describe the different steps to build metacells (Baran et al. 2019) from single-cell data using three frameworks: i. SuperCell (Bilous et al. 2022) (tutorial in 1.1), ii. Metacells version 2 (MC2 (Ben-Kiki et al. 2022)) (tutorial in 1.2), iii. SEACells (Persad et al. 2023) (tutorial in 1.3).

We provide examples of downstream analyses (section 3) performed at the metacell level (e.g clustering, differential analysis, marker identification).

We also show how to obtain metacells by running these methods using a command line tool that we provide as part of the MetacellAnalysisToolkit (MATK) github repository. This repository also contains the MetacellAnalysisToolkit R package which provides R functions to compute QC metrics and visualization functions to evaluate the quality of metacells.

In chapter 4, we use MATK to build metacells on a continuous dataset of CD34+ cells and use the R package to visualize the constructed metacells. The use of the MetacellAnalysisToolkit R package to visualize and evaluate the quality of the metacells is described in chapter 2 and 4.

Finally, using the MATK command line tool, we show in section 5 how we can use metacells to integrate and analyze around 580,000 cells from 107 individuals distributed in 166 samples.