Skip to main content
6 events
when toggle format what by license comment
Jul 22, 2020 at 14:50 comment added Genarito Perfect! Thank you!!
Jul 22, 2020 at 14:48 vote accept Genarito
Jul 22, 2020 at 14:38 comment added EdM @Genarito why not provide all the p-values (often "q" values for FDR) and let the user decide? The rank-ordering will be the same; the only difference among the criteria is how far down the list you go for a particular FDR. Provide links to the meaning and interpretation of the different criteria. That said, this pairwise correlation approach is throwing away so much information that I think it would be a much greater contribution to the field to provide a tool for analyzing such data in a way that takes advantage of the correlation structures in the data, like partial least squares.
Jul 22, 2020 at 0:26 comment added Genarito Case 1: gene expression with correlation vs. Mirna with correlation -> use Benjamini-Yekutieli method, adjusting with 3M p-values. Case 2: gene expression with correlation vs. Mirna without correlation -> BH adjusting with 3M p-values? Case 3: Same than previous one but miRNA dataset is the one with correlations Case 4: gene expression without correlation. MiRNA without correlation -> Which method would be convenient? Does the independency affect the number of used p-values to adjust? I really appreciate your help
Jul 22, 2020 at 0:26 comment added Genarito Thank you so much for your answer! You are giving a good insight into the solution. Considering that it will be a platform for multiple users, I would like to consider all the possible cases. Protein-coding mRNAs with correlation in expression patterns but also without correlation bacause the user applies some previous filter. The same for miRNA dataset. So, I would like to confirm that the election is the best in each case: (continues in next comment)
Jul 20, 2020 at 22:14 history answered EdM CC BY-SA 4.0