Hey there,

I'm a tad confused as to the best method to assess platform reliability after conducting a meta-analysis.

I've run both Pearson correlations on 50 common top genes across logFCs from all included data sets in a meta analysis as well as Bland-Altman. I understand that they assess very different things.

There are various conflicting studies on this with some using correlation to discuss agreement between microarray platforms and others the B-A.

The other confusing issue for me is whether I need them at all. If I'm assessing the top 50 genes or say even all of them from a meta analysis, can I truly talk about platform agreement here? Isn't biological sample variation and therefore measurement variation to be expected?

Thanks for any insights, pointers etc.!