Some of the highest throughput methods allow the capturing of hundreds of thousands of cells within a single experiment. These are usually accomplished using microcell or combinatorial indexing methods, or significant investment on commercial solutions like 10x Genomics Chromium. For plate-based methods, the limitations are usually on the cost and in physically handling many multi well plates. For droplet based approaches, it is often primarily an issue of cost.
Most often this question these days is asked in relation to 10x Genomics Chromium single cell captures. You can reasonably capture between a few hundred and 10,000 cells on a single Chromium capture lane, with the lower capture number resulting in a higher cost per single cell and the 10,000 target resulting in an expected 7.5%+ doublet rate (the probability that any given cell barcode has transcripts coming from two or more cells). Sample multiplexing strategies allow you to push this target capture number higher without compromising on doublet rate, but this only goes up to a limit as at the high end you start having to exclude a large proportion of your data as doublets (even though you can identify them as doublets, they still eat up reads). The Satija lab website has a nice tool for seeing how different factors play together: https://satijalab.org/costpercell. Important note here – in practice, the multiplexing doesn’t alway work out perfectly, so proceed with caution and manage risk (in terms of sample usage and experimental design).
An alternative to the approach of multiplexing and throwing out potential doublets is the concept of barcoding molecules prior to capture on droplet-based methods (like 10x). Not trivial to setup, but potentially a powerful and efficient hybrid approach: https://www.biorxiv.org/content/10.1101/2019.12.17.879304v1.full
Please login to submit your answer