Hi guys, maybe a very easy question but also a necessary one. I am conducting a RNA-Seq analysis, where I have to collect short read (bulk) RNAseq data from healthy mouse retina and brain tissues. Im currently struggling with finding studies that a have large amount of samples from these mouse tissues. I meanly used ArrayExpress and GEO databases to search for RNASeq data, but most of the studies I came across contained only around 3-6 (Wildtype) retina samples. Ideally I would want to collect the samples from one big studie, but thats unfortunately not the case due to the lack of studies/samples on this topic.
In order to have sufficient amount of RNASeq data , I collected samples from 3-4 different studies which I will use for downstream differential exon usage analysis. Since the samples are from different studies, it is expected that there will be somewhat biological/technical differences that could cause biases in the data. Also the read lengths in the fastq files are somewhat different. I suppose that normalisation and trimming of the data could deal with this bias problem, but I am not sure if this would be enough or are their other important things im missing?
I'd appreciate your advice on this topic! And if you might know any project/studies that have the data I am looking please let me know :)
Thanks in advance!
Yes, there will be big biases in the data, biases for which the causes can be known, and biases for which the casues will be unknown, but still there.
The extent to which it is possible to overcome these differences depend on both the design of the studies you are integrating and what you want to do with the data.
What is your downstream application? Are you looking to do differential analysis or network analysis or machine or learning or....
Differential analysis using DEXSeq is the goal to detect differential exon usage.
If you are comparing retina to brain, do at least some of your studies have both these, or do brain and retina samples come from different studies?
No the samples from the different tissues come from different studies. And within each tissue there are samples from different studies, but of course not al of them
If you don't have any studies that have both tissues, then there is not really anything you can do. You'll just have to accept that there is likely to be inter-study driven bias.