search

anders-biostat / MethSCAn

Python package with CLI for the analysis of single cell methylation data.
https://anders-biostat.github.io/MethSCAn/
GNU General Public License v3.0
11 stars 3 forks source link

batch effort remove #2

Open hengbingao opened 1 month ago

hengbingao commented 1 month ago

Hi anders,

Very powerful and effective tools for single cell methylation data analysis, but I wonder if you have consider the batch efforts across different library, and how to remove batch effort through MethSCAn. I think it will be nice if you could supply some strategy which could help to remove batch efforts based on MethSCAn.

Kinds regards, Hengbin

LKremer commented 1 month ago

Hi Hengbin,

this is a great question but unfortunately I don't have a satisfying answer. In principle it should be possible to adapt one of the many batch correction or integration methods from single-cell RNA-seq to single-cell methylation data. However I have never tried this, so I'm not sure how easy it would be. I will let you know if I ever find out how to do this! Lukas

hengbingao commented 4 weeks ago

Hi Lukas,

Thanks for your reply. In fact, I always consider if it is necessary for snMethyl data to remove batch efforts, becasuse for mC level, it always show 0-1 mC fraction and show same acrossing different sequencing depth library, it should be different bewteen RNA/ATAC/ChIP-seq. I wonder if it will remove the batch efforts after MethSCan smooth! Not sure but it should be a isssue which need to think over.

Hengbin

LKremer commented 3 weeks ago

Hi Hengbin, yes, I had exactly the same thoughts. In scRNA-seq batch effects are often so strong that you can easily see them by eye in e.g. PCA or UMAP. In single-cell methylome analysis I never saw strong obvious batch effects like that, but of course that doesn't mean that they are not there. Definitely an interesting topic to explore!

hengbingao commented 3 weeks ago

Hi Lukas, Yeah, I have tried to do that, I notice there are some publications have use the harmony to remove snMethyl batch efforts, I also successfully remove the batch efforts through harmony, but I found that harmony always be so strong to remove the batch efforts and it can not keep the samples difference, it is really annoying! So I think it should be better and more reasonable to consider a new methods which remove snMethyl batch efforts. Besdies, I always believe that if observed single cell mC of different snMethyl libraries show similar number/genome distribution/TSS enrichment.....,it should be more ideal to keep the sample difference without batch effort removing! I think snMethyl batch efforts maybe caused by library construction methods, only if we can check these single cell mC with different methodological library patteren, we can make sure if we need to remove batch efforts!