I want to compare result from my protein docking experiments to various protein structures of the same protein.

I have two samples and in Sample 1, I have 20 proteins, while in sample 2 I have 15 proteins.

Between two samples (1 and 2), 15 proteins are the same.

`**Sample 1**`                            `**Sample 2**`


  Protein (A)                               Protein (A)  

5                                           3
10 -- average (10)                          6 --average (6)  
15                                          9


Protein (B)                                 Protein (B)


43                                          11
46 -- average (46)                          13 --average (13)
49                                          15



and it goes up to Protein (T)               `there are 15 proteins in sample 2`         
so in sample 1 I have 20 proteins.

Importantly, each protein in two samples (Protein A for example) have 3 protein structures and the numbers for Protein A in sample 1 and for Protein A in sample 2 are not derived from the same protein structure but are derived from the same protein (A). Similarly for all other proteins.

In this particular case, what would be good scientific way (if possible) to statistically compare these two samples?