search

liulab-dfci / TRUST4

TCR and BCR assembly from RNA-seq data
MIT License
286 stars 50 forks source link

How to deal with sequencing depths in multiple samples. #218

Open orient100 opened 1 year ago

orient100 commented 1 year ago

Hi, thanks for providing the software. I have multiple samples and have run TRUST on each of them, but the sequencing depths are different between samples, do you think there is a suitable solution for this? Thanks

mourisl commented 1 year ago

The sequencing depth most affect your diversity measures. For example, I would imagine the Shannon entropy could be higher in deeper sequenced samples. But the number of TCRs/BCRs you can detect is also confounded by the lymphocyte infiltration level, so the final correlation is uncertain. You can examine whether the correlation holds in your samples by checking the correlation between diversity and the sequence depth.

For normalized diversity measure, such as clonality, I don't think the sequencing depth would affect the result very much.

WangKang-Leo commented 1 year ago

I would have similar concerns. If I detect significant batch effects among the population-based bulk RNA-seq data, should I correct this batch effect within final clonality tabular matrix?

mourisl commented 1 year ago

That's a good question. How do we correct the batch effect of clonality?