Table 2 The 14 RP features derived from DTI pair-specific one-to-all score curves.
From: Reciprocal perspective as a super learner improves drug-target interaction prediction (MUSDTI)
Feature generic name | Short name | Type | Description |
|---|---|---|---|
Y-in-X-percentile | ryx | Rank | Percentile of target Y among all the predictions for drug X |
X-in-Y-percentile | rxy | Rank | Percentile of drug X among all the predictions for target Y |
Adjusted reciprocal rank order | ARRO | Rank | Reciprocal product of rxy and ryx |
X-percentile-baseline | rxt | Rank | Percentile rank of the target nearest to the local cutoff value of drug X |
X-baseline | sxt | Score | Score at the local cutoff value of drug X |
Y-percentile-baseline | ryt | Rank | Percentile rank of the drug nearest to the local cutoff value of target Y |
Y-baseline | syt | Score | Score at the local cutoff value of target Y |
Percentile-difference-from-baseline-X | pdx | Fold | Difference between rxy and rxt |
Percentile-difference-from- baseline-Y | pdy | Fold | Difference between ryx and ryt |
Fold-difference-from-baseline-X | fdx | Fold | Fold-difference of target Y score in drug X from baseline sxt |
Fold-difference-from-baseline-Y | fdy | Fold | Fold-difference of drug X score in target Y from baseline syt |
SD-distance-from-mean-X | Stdx | Stats | The number of standard deviations from the mean score in drug X |
SD-distance-from-mean-Y | Stdy | Stats | The number of standard deviations from the mean score in target Y |
Original-Score | \(G<id>\) | Score | The original predicted score from the component model |
