Cell type annotation

Previous lectures…

Gene Counts

16

http://data-science-sequencing.github.io/Win2018/lectures/lecture16/ https://bioinformatics-core-shared-training.github.io/cruk-summer-school-2018

/SingleCell/slides/2018-07-25_CRUK_CI_summer_school-scRNAseq.pdf

cDNA alignment to Genome and group Results by cell

• In each gene, within each cell, the total number of unique UMI is counted and reported as the number of transcripts of that gene for a given cell.

^{Hundreds of millions of reads} Thousands of cells

Count unique UMIs for each gene in each cell

Create digital expression matrix

Gene counts matrix

Interpretation of zeroes

Ткани и органы - гетерогенны

Прочтение транскриптома группы гетерогенных клеток (bulk seq) может замаскировать существенные отличия в уровне транскрипции отдельного типа клеток.

На изображении справа - препарат тонкой кишки, содержащий десятки типов клеток.

https://www.pathologyoutlines.com/topic/smallbowelnormalhistology.html

What is a “cell type”?

• Originally defined in terms of function, location tissue type, cell morphology
• Later extended to
• presence/absence of cell surface markers
• gene expression (molecular profile)
• Currently very much less fixed
• cell cycle phase
• migration state
• differentiation: cell state

Why should we identify cell types?

• Samples are heterogeneous (in general)
• Tumor sample: how much do they differ from normal cell types?
• Find new cell types which have been missed by using

“standard” surface markers

• Follow cell fate and determine cell differentiation mechanisms
• To determine which cell types might communicate with each other
• To compare the abundance of cell types in different conditions*
• …

Change in cell type abundances: what are the new cells?

Mouse mesenteric lymphatic endothelial cells. From González-Loyola A et al 2021, DOI: 10.1126/sciadv.abf4335

Cluster ID

Wild type cells

TF-knockout cells

Cell surface markers

• Often considered the gold standard esp. in immunology
• mRNA of cell surface markers sometimes lowly expressed or absent
• Use a combination of such marker genes, and also other genes like marker genes among clusters (eg secreted proteins or transcription factors)

By Lokal_Profil, CC BY-SA 2.5, https://commons.wikimedia.org/w/index.php?curid=8797123

Markers of murine B cells and plasma cells. (Tellier J, et al., 2017)

Annotation methods

Manual vs automatic cell type annotation

• Manual: using marker genes
• What most people do…
• Time consuming
• Requires expert knowledge
• Sometimes subjective and inaccurate

​

• Automatic: requires a reference
• Use complete cell type-specific mRNA expression profiles based on bulk RNAseq from FACS-sorted ’pure’ populations
• OR: Use “a reference” of manually curated cells picked from scRNA-seq data sets
• Can miss cell types if they are not included in the reference
• Methods:
• Assign a cell type per individual cell or per cluster of cells (better per cell)
• Assignment of cell type via correlation of each cell/cluster to the

“reference”

Manual annotation using known marker genes

Microglia

Astrocytes

Human glioblastoma multiforme cells, 10x Genomics data (source of data to play with)

https://support.10xgenomics.com/single-cell-gene-expression/datasets/4.0.0/Parent_SC3v3_Human_Glioblastoma

Databases with cell type marker genes

• PanglaoDB https://panglaodb.se/ (mouse and human)

Check out https://cran.r-project.org/web/packages/rPanglaoDB/index.html

​

• CellMarker http://bio-bigdata.hrbmu.edu.cn/CellMarker/

​

• SingleR https://github.com/dviraran/SingleR (Aran et al. 2019), access via celldex package, eg human primary cell atlas (microarrays)
• Human Cell Atlas https://www.humancellatlas.org

(Regev et al) single cell RNA seq atlas, also some mouse data

• Single cell portal: https://singlecell.broadinstitute.org/single_cell
• And others (https://www.sciencedirect.com/science/article/pii/S2001037021000192)

Databases with cell type marker genes

SingleR

SingleR

Easy access to rich reference data:

• HPCA: manually-annotated Human Primary Cell Atlas 37 main types, 157 subtypes, 713 samples
• BluePrint +ENCODE

24 main types, 43 subtypes, 259 bulk RNAseq samples

- Mouse: ImmGen and ‘mouse.rnaseq’ (brain-specific)

Classifies cells to both main types and subtypes, performs both single cell-wise and cluster-wise annotation

Several methods are available

Abdelaal et al 2019 https://genomebiology.biomedcentral.com/articles/10.1186/s13059-019-1795-z

Check benchmarks!!

LLMs

Combine published scRNAseq datasets to create an atlas that can be used as a reference

ProjecTILs, an algorithm for reference atlas projection

Andreatta et al 2021 Nat. Comm. https://www.nature.com/articles/s41467-021-23324-4

https://github.com/carmonalab/ProjecTILs

Additional links

https://bioconductor.org/books/release/OSCA/cell-type- annotation.html#assigning-cell-labels-from-reference- data

​

​

Review on automated cell annotation, Pasquini et al 2021 https://www.sciencedirect.com/science/article/pii/S2001 037021000192

Slides adopted from https://sib-swiss.github.io/single-cell-training/day2/day2-4_cell_annotation.html