Agreement of functional annotation and expression data of genes. (a) Two groups of genes obtained after an a priori classification by, for example, functional annotation. The red lines connecting the elements within each group represent significant similarity between their expression profiles. The number of red lines with respect to all possible ones gives and indication of the consistency of the group. Thus, group a is very consistent, as all the possible connections but one, a3-a4, are made. On the other hand, the consistency of group b is poorer, as its elements form two subgroups (b1-b2 and b3-b4). The blue lines connecting the elements between each group are relations lost upon the a priori classification used. The higher the number of lost connections the less comprehensive the classification will be. In this case, the subgroup b1-b2 significantly correlates with group a. (b) Consistency of the functional groups established by gene annotation. A decreasing trend implies loss of interactions between members. F--, -P-, FP-, F-L, -PL and FPL indicate functional classes as defined in the text. For example, class F-- contains all gene pairs in which both members have the same general function regardless of their pathway/system (P) or location (L), class FP- contains all gene pairs with the same general function and the same pathway/system but not necessarily the same location, and so on. (c) Comprehensiveness of significant gene-expression pairs in the functional groups established by gene annotation. The increasing trend suggests that genes correlating at high CCF values tend to belong to the same functional class. This is especially obvious when a broad functional classification is used in which nearly all the possible pairs in the experiment are represented at high thresholds. The percentage of the gene pairs in each group with respect to the total number of pairs was: ALL (100%), F-- (8.8%), -P- (1.2%), FP- (4%), F-L (6.7%), -PL (1%) and FPL (3.1%). The group --- was not included in the -P- and -PL classes.