Schematic view of the CORRIE annotation framework. The only requirement for CORRIE is a classification of sequences. Here, we start with the classification of enzymes found in SwissProt. This enables us to create two tables, one for sequences and one for classes. From pairwise sequence comparisons we derive a score table, which describes all the classes hit by each sequence. BLAST scores are further integrated into correspondence indicators (CIs), which describe the relationship each sequence has with the classes it hits. Next, CIs are integrated to compute the probability that a sequence belongs to a particular class. The table "CI reference" is central to the framework as it constitutes a reference against which new proteins are compared and classified. This is illustrated in Figure 2.