I am looking to perform a cumulative analysis of data from multiple RNASeq studies deposited online. Some of the studies uploaded the raw fastq files, but others only upload a post raw mapping counts table or normalized counts ex) https://zenodo.org/record/4114617
I do not have the computing power to perform the same HISAT2 alignment for the other studies raw fastq to compare (normally use salmon). There is an additional in study I want to compare to that uses a third alignment method and only uploaded a raw mapped table.
My end questions are can I compare these different mapping methods on any level and account for it statistically? I know I can run hisat2 and let it run for ~1 week + but I would lose the third group even if possible.
Finally, assuming I ran a hisat2 alignment and quantification from the raw fastq files of study 1 to compare to the hisat of study 2, would I be to compare or swap the controls of each study? The reason is many of these “controls” aren’t true controls and am interested in determining control sample variability.
Thanks,
Luke
no.
The best you can do is to draw conclusion using one dataset, and validate your conclusion using other datasets.