NTB-T04

Title

Functional analysis of single-cell transcriptomics

Tutorial details
  • Date: Sunday, September 18th
  • Time: 09:00 to 13:00 CEST (Slot 22)
  • Format: Face-to-face
  • Room: TBD
Instructors
  • Pau Badia i Mompel, PhD candidate, Saez-Rodrigez Group; Heidelberg University Hospital, Institute for Computational Biomedicine, Bioquant, Heidelberg (Germany)
  • Robin Browaeys, Data Mining and Modeling for Biomedicine, VIB Center for Inflammation Research, Ghent, (Belgium)
  • Daniel Dimitrov, PhD Candidate, Saez-Rodriguez Group; Heidelberg University Hospital, Institute for Computational Biomedicine, Bioquant, Heidelberg (Germany)
Summary

Recent advances in omics technologies have led to a rapid increase in the popularity and applications of single-cell data-sets. Standard analyses and workflows solely focus on basic preprocessing steps followed by the identification of differentially expressed genes, and their subsequent use in cell-type annotation and characterization of biological processes. In this tutorial, we show how prior knowledge can be used to extend each of the aforementioned steps, as well as to extract clear biological insights. Furthermore, we provide an introduction to the state-of-the-art intercellular communication methods, as tools for systems-level hypothesis generation tools in single-cell data. We thus cover a diverse set of prior knowledge resources and show how these can be used to support and extend analysis steps ranging from quality control, cell-type annotation and transcription factor and cytokine activity inference. Finally, we show how advanced functional omics analyses can be used to refine cell-cell communication predictions.

Intended audience

Broad audience with some experience in omics data-analysis and interest in analyzing single-cell transcriptomics.

Prerequisites
Maximum number of attendees

30

Material required (for participants)

Laptop with at least 4GB of RAM memory and 10 GB of free disk space, with the package management system Conda installed.

Programme
  • Brief summary to single-cell transcriptomics pre-processing (25 min)

To start, we will briefly go over the basics of single-cell transcriptomics: reading raw data, QC filtering, normalization, feature selection, dimensionality reduction and visualization using the scanpy framework

  • Prior-knowledge-driven cell-type annotation (20 min)

We will introduce the concept of gene sets, and how to use them to aid in cell identity annotation using over-representation analysis and PanglaoDB, a database of cell type-specific marker genes.

  • Gene set enrichment of biological terms (20 min)

Then we will show different gene set data-base resources like MSigDB, and how they can be coupled to a statistical method like GSEA or over-representation analysis to infer enrichment of biological terms in scRNA-seq.

  • Transcription factor and cytokine activity estimation from prior-knowledge footprints (30 min)

Afterwards, we will introduce the concept of footprints and compare them to gene sets. As examples, we will show how to infer transcription factors and cytokine activities with the decoupleR framework, making use of the gene regulatory network DoRothEA and CytoSig’s cytokine signatures, respectively.

  • Break (15 min)
  • Cell-cell communication inference (30 min)

We will introduce the audience to the relevance, concepts, and assumptions behind intercellular communication inference. We will showcase these with a hands-on tutorial on ligand-receptor interaction prediction methods, including CellPhoneDB, NATMI, and SingleCellSignalR, via LIANA.

  • Modeling intercellular communication by linking ligands to target genes (25 minutes)

We will then show how one could go beyond ligand-receptor interactions by incorporating prior-knowledge of intracellular signalling. This part will largely focus on introducing the audience to the NicheNet method.

  • Connecting biological activities to intercellular communication (15 min)

Finally, we will summarize the lessons from the tutorial by connecting the results from different analyses, and thus highlighting their potential as complementary hypothesis-generation tools.

TIME CONTENT
9:00 – 9:25Brief summary to single-cell transcriptomics pre-processing
To start, we will briefly go over the basics of single-cell transcriptomics: reading raw data, QC filtering, normalization, feature selection, dimensionality reduction and visualization using the scanpy framework
9:25 – 9:45Prior-knowledge-driven cell-type annotation
We will introduce the concept of gene sets, and how to use them to aid in cell identity annotation using over-representation analysis and PanglaoDB, a database of cell type-specific marker genes.
9:45 – 10:05Gene set enrichment of biological terms
Then we will show different gene set data-base resources like MSigDB, and how they can be coupled to a statistical method like GSEA or over-representation analysis to infer enrichment of biological terms in scRNA-seq.
10:05 – 10:35Transcription factor and cytokine activity estimation from prior-knowledge footprints
Afterwards, we will introduce the concept of footprints and compare them to gene sets. As examples, we will show how to infer transcription factors and cytokine activities with the decoupleR framework, making use of the gene regulatory network DoRothEA and CytoSig’s cytokine signatures, respectively.
Break
10:50 – 11:20Cell-cell communication inference
We will introduce the audience to the relevance, concepts, and assumptions behind intercellular communication inference. We will showcase these with a hands-on tutorial on ligand-receptor interaction prediction methods, including CellPhoneDB, NATMI, and SingleCellSignalR, via LIANA.
11:20 – 11:45Modeling intercellular communication by linking ligands to target genes
We will then show how one could go beyond ligand-receptor interactions by incorporating prior-knowledge of intracellular signalling. This part will largely focus on introducing the audience to the NicheNet method.
11:45 – 12:00Connecting biological activities to intercellular communication
Finally, we will summarize the lessons from the tutorial by connecting the results from different analyses, and thus highlighting their potential as complementary hypothesis-generation tools.
Keep me updated