Correspondence of DPA predictions to annotations.^a

aEach entry in the table contains a number of DPA predictions that have a given type of annotation. Percentages are with respect to the numbers in the four subsets of DPA predictions indicated at the head of each of column. The subsets are increasingly restrictive from left to right: predictions in a domain for which a transfer MSA is available; predictions in a domain for which a conservation MSA is available; predictions for which the conservation P-value is no greater than 10−2; and predictions for which the conservation P-value is no greater than 10−3. Annotations are organized into three major sections that are increasingly expansive from top to bottom, as described in the text: direct annotations, protein-level annotations, and family-level annotations. The meaning of the NSM, NSM-valid, CSA, and Text annotations is as described in the text.