Best suitable statistics

Hi guys, I'm trying to see if different treatments can be related (or similars), although they are independent and unrelated treatments.
Since I'm working with a parental toxin and some of its metabolites (which were used in distinct experiments), i was planning to relate them somehow and make statements such as "the response of the metabolite X after 6h is 23% similar to the response of the parental toxin after 12h". That could be helpful to assess the temporal effects related to the metabolism.
I tried Pearson, but it's not that correct, because then I'd be measuring similarity (since they are independent). I tried linear regression, and multiple linear regression (using the metabolite treatments that showed the highest pearson coefficient as independent variables), but I'm still in doubt. Can someone suggest something?

PS: I used correlations, binary correlations, tried to use the slope coefficient as measure (and so I had two regression lines - since I don't have independent and dependent variables per se), but I still cant' figure it out how to relate then.