Selecting Negative Samples for PPI Prediction Using Hierarchical Clustering Methodology

Table 2

New sizes of datasets after filtering process.

Datasets	Size of filtering training set with GSN set obtained using the presented hierarchical clustering	Size of filtering training set with randomly selected GSN set	Size of filtering training set with “balanced’’ GSN set obtained from the approach by Yu et al. [38]
Binary-GS	933	937	987
Ito-core	680	686	700
LC-multiple	2362	2380	2468
Uetz-screen	574	584	594
Random negative dataset 1	4893	4894	4894
Random negative dataset 2	4895	4894	4898
𝑅^test_13	4735	4995	4992
𝑅^test_23	4788	4995	4994
𝑅^test_33	4814	4991	4991
𝑅^test_43	4844	4987	4992
𝑅^test_53	4854	4983	4986
𝑅^test_63	4816	4991	4994
𝑅^test_73	4837	4985	4990
𝑅^test_83	4797	4994	4994
𝑅^test_93	4873	4994	4996