Glue dependence
The headline view. Δ-ipTM against bridge percentile, coloured by experimental label, sized by the number of lysines in the peptide. Lasso or box select to push a set into the pinned selection; every other tab follows. Δ-ipTM is the spec's definition (minimum across seeds); the median is in every tooltip, because one collapsed seed can invert the minimum.
Matched-control ROC
Δ-ipTM on the Sievers degraded versus non-degraded split. The negatives came out of the same screen as the positives, which is what makes this curve mean anything. Drag the threshold to read precision and recall off the live numbers.
Not computable yet
The ROC needs the Sievers matched set, which the campaign runs after the benchmark panel. With only benchmark pairs scored there are no matched experimental negatives to plot, and a curve drawn without them would be meaningless rather than preliminary.