You are viewing a single comment's thread from:
RE: Lab Diaries #4 - Whole Transcriptome and Proteome Analysis by RNA Sequencing and TMT-MS
I'm curious if you're using other statistical tests for differential expression? Deseq2 comes to mind. I'm wondering that if you used that (or adjusted alpha for you p-value test multiple testing) if it might not cut down the initial number of genes and proteins to look at.
I'm familiar with RNA-seq but have no proteomics experience. How big (in terms of memory) do those datasets get?
Thank you for your comment! Basically, DESeq2 would not improve analysis from the biological point of view that much, improvement may be visible in those samples that have large deviation caused by measuring error only...
To obtain biological meaning of datasets, I use GSEA which is based on Kolmogorov–Smirnov algorithm, and then I analyse biological pathways and genes within them.
For proteomics data, size is not even close as of RNA-seq - you get your results in one Excel file :)
That makes sense. I arrived at DESeq2 through examining microbial population changes and it's been super helpful in that context. It's always fun to see how a tool of choice can vary in efficacy between problems.
As I'm doing the analyses in biology and chemistry I understand your standing point.
In almost all the cases, we are able to offer them "better" math, but... Those things are alive.
We often start with bad data or we get qualitatively the same thing at the end.
I also noticed that sometimes we have "one-time hit" after application of exotic math that works well, but only once, on one dataset.
Due to this, if the procedure works for the data (signals) - don't touch it, you will break something :)
If they are stuck with data we are coming to the rescue.