search

jlakkis / CarDEC

Other
19 stars 4 forks source link

Applying CarDEC to bulk RNA seq #8

Open mdanb opened 3 years ago

mdanb commented 3 years ago

Is CarDEC applicable to bulk RNA seq data? I have a relatively large dataset (~1300 samples from 70 experiments) that I’d like to correct for significant batch effects

eleozzr commented 3 years ago

I think our CarDEC probably could be applied to bulk data. You can first try the z-score step across all samples, since there may not be enough bulk samples per experiment to justify within batch normalization.

But batch-specific normalization is important for batch effect removal, so maybe you should normalize your data across within each experiment if there are at least 3 samples in each experiment.

mdanb commented 3 years ago

@eleozzr would you be interested in helping me adapt it to bulk RNA seq? I have a limited amount of time and I'm writing a paper. If you can help out, I'll add you as an author, regardless of whether or not it ends up helping to adjust the batch effect. Let me know whether or not you're interested by emailing me: mdanb@umich.edu

mdanb commented 3 years ago

@jlakkis just want to mention you also to see if you're interested. Again, please email me to let me know (whether you're interested or not).

mdanb commented 3 years ago

Also, @eleozzr regarding your reply, another question is does it even make sense to speak of "HVGs" when samples per experiment/batch in bulk RNA-seq are usually low compared to samples per experiment in scRNA-seq? Would appreciate your input on this!