A query about theoretical total input of spike-ins about sc_mixology HOT 2 CLOSED

Hi Wenhao,

I have added the ERCC annotation: https://github.com/LuyiTian/CellBench_data/tree/master/misc.

It should be used with care though. We have noticed the calculated capture efficiency is much lower than we expected and we thought it might be caused by ERCC degradation because there are a couple of freeze-thaw cycles and dilution. Also, we used a very high dilution factor, so as you might see many ERCC genes have a theoretical input lower than one, so cannot be detected even with 100% capture efficiency. personally, I found the methods that use ERCC to separate technical variation do not perform very well, such as BASiCS.

The last thing is that the RNA mixture data were generated from pure RNA instead of cells, where ERCC spike-in behaved well. But the ERCC have more variation in real single-cell data, which might be related to cellular content, as ERCC have short polyA and no secondary structures or attached proteins, making them more vulnerable than cellular mRNA.

from sc_mixology.

Hi @LuyiTian, I see, thank you so much!

Best wishes,
Wenhao

from sc_mixology.