I have a problem finding a methodology to compare proteomic profiles against another. The experiment is the following:
Experiment 1. Ten pdx tumour models with their proteomic profiles were done in one Research Institute with DIA-MS. All of them are the same tumour from different patients. Differential analysis was done with Limma.
Experiment 2. Same tumour, but this time entirely on humans (not pdx) and also controls were included (SWATH-MS in another research institute). Differential analysis was done by t-test.
The biological question is which of the 10 pdx models resembles better with the human one.
I tried different ways of normalizing together the raw data but there is a huge difference in the levels of abundance between both experiments. Also, the pdx ones are really similar between them.
In the end, I did a one vs rest contrast on the pdx conditions to get the unique markers for each model and use them as genesets (positive and negative separated) to try to see if the upregulated and downregulated genes (tumour vs control on experiment 2) are some kind of enriched (pre-ranked GSEA with rank as -log10 p-value*log2FC).
Do you think this can be a solution? Is there a more simple way to do this assessment?
Thank you!!
