Table 3 Description of DR case subsets and rest of possible combinations of proteins in DISNET. The total number of possible combinations of protein pairs is 171,152,751
From: Protein sequence analysis in the context of drug repurposing
Source of DR information | Subset | DR vs DISNET cosine distance protein pairs p-value (mean ± standard deviation) | |||
|---|---|---|---|---|---|
OneHot | SGT | ProtBERT | SeqVec | ||
RepoDB | Unfiltered | 0.00 (0.0947 ± 0.0597) | 0.00 (0.2446 ± 0.1055) | 0.00 (0.5288 ± 0.2396) | 0.00 (0.5246 ± 0.1415) |
Filtered | 4.30e-07 (0.0793 ± 0.036) | 1.66e-14 (0.1797 ± 0.0893) | 0.01 (0.4966 ± 0.2477) | 1.80e-03 (0.4875 ± 0.1498) | |
Literature | Unfiltered | 0.00 (0.0942 ± 0.0609) | 0.00 (0.2574 ± 0.1058) | 0.00 (0.5411 ± 0.2516) | 0.00 (0.51819 ± 0.1421) |
Filtered | 2.65e-13 (0.0593 ± 0.0290) | 1.65e-11 (0.1778 ± 0.0783) | 6.715e-05 (0.4436 ± 0.2243) | 6.44e-06 (0.4436 ± 0.2243) | |