Fig. 1
From: A close look at protein function prediction evaluation protocols
Overview of the NA and NP setups. We distinguish between three sets of annotations that are used to define the training and test set in the two setups. Annotations accumulated between an initial time t 0 until t 1 (end of 2009 in our experiments) and form a set A, which is the training set in both NA and NP. The set of annotations acquired for those proteins after t 1 form a set B, which is the test set in the NA setup. The set of annotations acquired after t 1 for proteins that were unannotated before t 1 is denoted by the set C, and is used as the test set in the NP setup