Hi everone,
I have been involved in a work where we have designed a "Generative Adversarial Network" . The task of this network is to produce realistic enhancer sequences of the human as it is trained in a very generic way on FANTOM5 enhancer dataset. The model is generating sequences and are being classified as enhancers (by our own classifier) as well, but still we are not clear that how can we validate our results.
Real Enhancers -> GAN Model -> Learning -> Synthetic enhancers -> ?????
Anybody knows how can we actually validate them ? Any kind of search engine, methodology or any literature would be helpful from your side
Before answering, is your group exclusively in silico or do you have access to a standard molecular lab and tissue culture? Any meaningful validation will involve experiments imho, therefore I am asking.