What are the considerations for deciding how deeply to sequence a scRNA-Seq sample?
For droplet-based end-counting libraries we find that about 50,000 reads will cover most use cases for data analysis. You achieve surprising decent separation of diverse cell types even with as low as 5,000 reads per cell (see attached example), but something about 20,000 per cell is more reasonable. If you are trying to discriminate difference between more fine-grained subtypes, more reads should help. Also, if you are looking to run something like a variant detection or something like velocity inference, the additional reads will also help.
Minimum Sequencing depth for any type of Library:
20,000 read pairs/ cell
Immune profiling (VDJ)
5,000 read pairs/ targeted cell
Gene Expression with Feature Barcoding technology
Minimum 5,000 read pairs/cell
25,000 read pairs per nucleus (50,000 individual reads. 25,000 from R1, 25,000 from R2
750,000 read pairs per cell (for human) enables accurate detection of 2 Mb events per single cell
Please login to submit your answer