Single cell and spatial omics: A short introduction to the core concepts of scRNA-seq and more

A short introduction to the core concepts on single-cell omics data and spatial omics data. I will start by introducing how these types of data relate to and differ from normal omics data and concisely explain the typical experimental workflows used to produce such data. I will then talk about some of the computational aspects of analyzing single-cell data, including pooling of cells, clustering to identify cell types, and the concept of pseudo-time. Finally, I will briefly talk about how analysis of spatial omics data relate to analysis of images.
0:00 Introduction: reminder of omics is and introduction to single-cell and spatial data
0:42 Single-cell workflow: dissociated cell culture, isolation of single cells, omics on single cells, UMIs, and multiplexing
2:24 Spatial workflow: in situ capture, spatial indexing, spatial RNA-seq, single-cell resolution, and laser capture microdissection
3:48 Computational analysis: single cell vs. bulk data, cells vs. samples, pooling, cell-type clustering, pseudo-time reconstruction, and analysis of spatial patterns

[@sara_rojo]

I watched several videos trying to find an easy explanation about this. You made it so simple. Thank you. Best regards from México:)

[@larsjuhljensen]

Glad it was helpful!

Great introduction thank you.

[@larsjuhljensen]

You are welcome!

[@geparada88]

Wow, such cutting-edge topics you are bringing to this channel. This is definitely one of my favorite areas of bioinformatics. There is so much more to say about scRNA-seq, but your video offers a nice and concise introduction. I very much agree that the sparsity of scRNA-seq data can be overcome by generating pseudo-bulks. In my case, I do clustering, followed by cell assignment and random pseudo-bulking. In this way, I artificially enhance the depth of my measuring units, while I can keep the cell type-specific information :)

[@larsjuhljensen]

Thanks a lot! What we're doing a lot is not to assign cell types, but rather to use variational autoencoders to compress the cells into a lower-dimensional latent space. It accomplishes much the same, since compressing the data into a lower-dimensional space inherently means that you are in some way averaging/summing up cells.

[@geparada88]

@@larsjuhljensen That sounds really interesting. I've been thinking along these lines too but in a less sophisticated way (just using knn to gather cells for pseudo-pooling across a lower dimensional space).
I would need to learn about autoencoders to really understand your strategy, but I kind of get the point.

[@larsjuhljensen]

You can read about it on bioRxiv: doi.org/10.1101/2022.07.06.499022
Or you can watch Mikaela from my group explain it: czcams.com/video/XyfsK4oujVc/video.html

[@mohammadghorbani3003]

That was amazing sir. You teaching bioinformatics and sharing your knowledge of omics, is the best course that one can take.

[@muhammadwaqasluqman2977]

wow, thank you so much.

[@larsjuhljensen]

You're welcome, I hope this short overview was useful!

[@pancake9191]

how am I supposed to understand without even one illustration picture

[@larsjuhljensen]

I would have loved to have good, simple graphical illustrations of the many different assays too, but the figures I could find were overly complicated and not suitable for a short overview presentation.