Figure 2
From: CORRIE: enzyme sequence annotation with confidence estimates
Illustration of the probability calculation implemented in CORRIE. To annotate a new sequence s, s is first aligned against all proteins in CORRIE. Here, s has similarity with proteins from two distinct classes: A and B. CIs between s and A, and between s and B are calculated [10]. The probability that s belongs to A (i.e. that s has function A) is calculated by comparison of the CI between s and A, with the CIs of proteins that belong or not to A. In this case, the ten proteins closest to s in the CI space are shown in the red dotted rectangle. Since all ten proteins truly belong to A, CORRIE estimates to P = 1 the probability for s to truly belong to A. When considering class B, ten proteins closest to s in the CI space do not belong to B. Therefore, CORRIE estimates to P = 0 the probability for s to truly belong to B. In this case, s would be annotated as having function A with probability 1.