Figure 7
From: High-precision high-coverage functional inference from integrated data sources
MW decision rule is the end result of carefully tuning an adjustable decision rule. The adjustable decision rule (equation 3) is tuned to optimize function-annotation performance for the integrated FLN by empirically testing a range of alpha values. MR, NW, and MW rules are special cases derived from this adjustable rule, with alpha set to be 0, 1, and infinity, respectively. When alpha is equal or above 10, the optimal performance is obtained, which is approximates to the performance of MW. The performances are evaluated by the annotation precision-coverage curves.