I want to perform a differential expression analysis by using data obtained from the new RJunBase database (http://www.rjunbase.org/rJunBase/download/toIndex). However, I am a beginner to this and I am having some trouble regarding the initial processing steps.
I have read their paper (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7779070/) but I am a bit confused with the kind of data it is provided (I have downloaded the linear junctions file "CHOL_expr.tar.gz"). The paper says the formula is: raw_counts/sum_counts × 10 000 000 (sum_counts represent the sum of the counts of all samples). However, as it is based on the ASJA algorithm (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6709372/), I do not understand which expression units they are (TPM, CPM...).
I have analysed the data, and it seems that it can't be comparable between samples (I have read that you can't compare samples if the total sum of reads is different?). I do not know if it is suitable for any specific package to perform the differential junction usage or if I should log transform it and perform any statistical test (as done in the clustering Figure of the ASJA paper). Has somebody worked with this new database or with this ASJA algorithm?
Thanks for your reply!
