search

statgen / demuxlet

Genetic multiplexing of barcoded single cell RNA-seq
Apache License 2.0
116 stars 25 forks source link

Too many doublet #109

Closed kxxxjo closed 4 weeks ago

kxxxjo commented 3 months ago

Dear All,

Hi I'm KJ from Korea. First of all, thanks for developing amazing demultiplexing tool.

I'm currently trying to demultiplex 2 and 3 samples using your demuxlet tool, but I'm encountering an issue regarding doublets. When we ran demuxlet, the output showed a high number of doublets: image

As shown in the image above, our results detected an excessive number of doublets. However, we have verified that our experimental steps for preparing GEMs are accurate and free of issues.

Could you please advise us on why we are getting these results and how we can resolve this issue?

Thank you for your assistance.

Best, KJ

hyunminkang commented 3 months ago

Try looking at the distribution of singlets and if the distribution of matched singlets look reasonable. If genotype is incorrect for some individuals, you will end up having too many doublets.

kxxxjo commented 3 months ago

Thanks for the quick reply :)

We have sequenced our WGS with 30x depth. I think 30x is enough for detection of SNP. Are you saying the WGS data is wrong?

Thanks!

Best, KJ

hyunminkang commented 3 months ago

I would suspect that (1) there might be sample swaps, or (2) your genotypes are not stringently filtered. Make sure to use only verified SNPs that are found in 1000G (no indels), with low missingness. It is common that using WGS-based SNP calls is a cause of problems if you do not filter genotypes carefully.

kxxxjo commented 3 months ago

Thanks for your useful feedbacks!

We will let you know after trying based on your comments.

Thanks.

Best, KJ