Image was produced with Midjourney model version 6.1. Prompt was 'abstract painting of small intestine made of small paint spots'.

Corresponding lectures

Lecture 12 - Single cell RNA-seq – principles and processing Lecture 13 - Analysis of scRNA-seq data using R Lecture 14 - Advanced topics in single cell genomics R

Description

You have now been introduced to the basics of scRNA seq analysis including how to carry out read mapping and barcode assignment, and how to read these data into R and perform quality control. We also explored how to normalize data and generate clusters and UMAPS which represent the data in 2D space. Finally, you integrated multiple different samples and performed rudimentary differential expression analyses and cell type annotation. In the lectures for this week, you were introduced with how to handle one type of multimodal data, CITE-seq, including integration and Weighted Nearest Neighbor analysis to generate clusters and UMAPS that captures the infleunce of both modalities on the structure of the data. The analysis was expanded to include how to subcluster a population of interest for downstream analyses.

In this lab, you’ll use the skills you’ve gained to explore a large unpublished study from the ‘MIST’ program, an NIH/NIAID-supported group of infectious disease and mucosal immunology researchers.

What you’ll need to do

To get started, download this seurat object. This is downsampled scRNA-seq data from the distal most portion of the small intestine (ileum) from mice. This may sound familiar from lab last week. The data is very similar, but we’ve included cell type annotations and the antibody (ADT) assay data.

Of course, you’ll also need to download the course DIY_scRNAseq_advanced.R script

Part 1

Read the Seurat object into a clean R environment. The seurat object contains two assays, RNA and ADT, and both have already been integrated for you.

  • Plot the integrated UMAP based on RNA and the integrated UMAP based on ADT
  • Perform WNN analysis (FindMultiModalNeighbors()), find clusters, and create a UMAP using both modalities. Plot the UMAP

Tips

  • Explore the content of the Seurat object first. What UMAPs (reductions) are available already?
  • 30 PC dimensions were used on the RNA and 20 on the ADT - this information is useful for the dims.list parameter in the FindMultiModalNeighbors function
  • You should not need to change the reduction list (reduction.list = list("integrated.rpca", "integrated.ADT.rpca")) when calling FindMultiModalNeighbors

Part 2

Cell types have already been defined for you at both a low (CoarseLabels) and high (FineLabels) level of granularity. Plot the frequencies of each coarse cell type across the different infections.

Tips

  • you will need to use “CoarseLabels” and “InfectionStatus” in setting up your clusters and conditions
  • unique(seurat_obj$InfectionStatus) will tell you which infections are in the data. Remember to use these as the “levels” when factoring your data before plotting the frequencies

Part 3

Subset a coarse cell type of your choosing and perform the steps to recluster and generate a new UMAP with this data as in the lecture.

Tips

  • There are not many enteric nervous cells - do not subset and recluster these
  • Memory management: when you finish this section, you might want to delete the subsetted seurat object from your environment.