A question about simulating multi-omic data from single-omic data

[HelloWorldLTY]

Hi, I intend to clarify my understanding about this function, shown in this tutorial:

https://songdongyuan1994.github.io/scDesign3/docs/articles/scDesign3-multiomics-vignette.html

It seems that we still need paired multi-omic data to estimate two sets of parameters, and then we can simulate arbitary omic data based on the joint parameters. Is my understanding correct? Is it possible for me to directly simulate one omic data from the other omic (say, based on scRNA-seq data to simualte scATAC-seq data)? Thanks.

[SONGDONGYUAN1994]

Hi,
Yes, I think your understanding is correct. Simulating one omic data from the other comic is not doable in scDesign3 --- that is almost like predicting one omic from another omic, which is not that easy. One idea is that you get the parameters from your scRNA-seq data (gene mean/variance/correlation in each cell type) and convert it into the parameters of scATAC (by gene - regulatory element relationship, I guess?) and use different distribution (ATAC is much more sparse). I haven't investigated this before and I would like to listen to your idea.

Best,
Dongyuan

[HelloWorldLTY]

Ok, I see, thanks.