Figure 5.

Employed datasets. For each combination C_i of known and unknown cell compartments ten datasets were created. They enable five-fold cross-validation with respect to each type of protein location – known ($k_i^j$) and unknown ($u_i^j$).