I know that in an ideal experiment, one library prep method should be used for all samples.
But I would like to know if it would be possible to use 4 library prep methods/kits on my samples and then send them for sequencing? What differences would there be between the different preparation methods? How would I account for these differences during the bioinformatics analyses?
My samples have varying RIN numbers and concentrations of RNA and so using just one library preparation method/kit would be ideal for some samples, but no good for the other samples, hence why I would need to use 4 methods.
To answer a few of your questions; 1) The RNA was obtained from primary skin cancer samples (Hence the variation in quality and quantity) 2) All library prep methods would produce the same sized libraries for sequencing 3) All the samples are the same type/kind >> Skin cancer_1 marker positive (SC1pos) & Skin cancer_1 marker negative (SC1neg). So far, I have 9 samples (18 total). 4) For the bioinformatics analyses, I would like to generate DEGs for the skin cancer marker positive population (SCpos_vs_SCneg)
Thank you very much in advance!
Hmm, yes thats that I was thinking but I have not worked with library prep methods personally so am not sure (Usually the company that I am doing sequencing with sorts all of that stuff out). I will contact them and ask if this is feasible with my samples. Otherwise I guess I will have to sacrifice samples and stick to one prep method. Thank you for your suggestion!
Absolutely try to go with just one method. The library prep step is one of the key steps that has been shown over and over again to introduce significant bias depending on the method, and no bioinformatics magic will be able to completely rescue that.
If one method is not feasible, the only way to move forward is to make sure that the library preparation protocols are not confounded with your contrast of interest, i.e. for each library prep method you should have multiple replicates from each condition.
You will get strong batch effects due to kit. Identical samples produced with different kits was what I used in this tutorial linked below as a starting point, maybe it helps to get you familiar plus some code suggestions for batch correction, but in general you should, as Friederike says, strictly stick to one kit if possible, and if not then at least make sure that each experimental group has replicates with each kit to allow batch correction.
Basic normalization, batch correction and visualization of RNA-seq data