Applications
Sten Linnarsson’s lab released two pre-prints on cell atlases (yes, not one atlas but two atlases) of the developing human brain and the adult human brain. I learnt a ton from the Methods alone and I bet you will too.
Methods
100x your single-cell analysis speed by using GPU-based NVIDIA RAPIDS. Here is a walkthrough on modifying your existing code to use RAPIDS implementation.
Mo Lotfollahi presented mvTCR, a multi-view deep learning model to integrate TCR sequences and transcriptome.12 mvTCR allows the distinction of antigen-activated T cells from bystanders. It also scales to millions of cells.
Conversations
What is your favourite single-cell data integration methods? In this thread, many people shared their thoughts, and as always, there is no one true algorithm. scVI seemed to be the favourite of many, though harmony was also mentioned a lot. However, stupidly simple methods, such as downsampling UMI counts, could actually be the most robust option.3 Anyway, I would recommend to focus less on batch removal (please stop using UMAP or tSNE as your main evidence of a ‘successful data integration’, whatever that means) and emphasise preservation of biological variance. The scib pipeline can help you with measuring batch correction and biological variance conservation metrics.
Here’s a thread4 sharing DropletQC, a great method to separate cells and empty droplets by Joseph Powell’s lab. Python users, rejoice! Apparently, Sten Linnarson provides an implementation to do this in Python using his lab’s cytograph-shoji.
Miscellaneous
Learning Resources
Do you want to get started in single-cell proteomics? Prof. Nikolai Slavov is here to help. Here is the link to his talk. His lab website also provides a plethora of information on aything related to single-cell proteomics.
Mark Sanborn (@Sanbomics) provides numerous single-cell RNAseq analysis live-coding videos in his Youtube channel. His complete single-cell RNAseq analysis walkthrough is particularly great.
Acknowledgements
Credits to the original authors of the Twitter threads. I am here only to curate stuffs :)