A node in the network is a random variable that can have one of two values, *false* or *true* (0 or 1, respectively).

Both the brightness and the size of a node represent the strength of the probability that the corresponding molecule is *present/active* in the tissue or cell of interest, *P*(*Vi* = 1).

A higher probability is depicted with a lighter color and larger ball radius (see key to the node color and size); when the *P*(*Vi* = 1) drops to 0, the node disappears from the figure (the ball radius drops to zero).

Each arc is a random variable with three possible different values: *inhibit, activate*, and *no effect* (*−1*, 1, and 0, respectively).

Complete confidence that an arc *A*_{V,U} represents an inhibiting function (*P*(*A*_{V,U} = −1) = 1) would be drawn as a thick bright-blue edge with a disk at the end (the leftmost edge in the key to the figure).

If both probabilities (*P*(*A*_{V,U} = −1) and *P*(*A*_{V,U} = 1)) drop to zero (indicating that *P*(*A*_{V,U} = 0) = 1), then the edge vanishes from the figure, indicating the *no effect* value.

(A) An internally consistent set of prior probabilities.

The resulting marginal distributions are either unchanged (on the input nodes G and B and on the sink node E) or have a decreased entropy (on all arcs and on nodes C and D), in contrast to the prior probabilities.

(B) An example with inconsistent prior probabilities.

The marginal distribution for arc *A*_{BC} is reversed with respect to the prior.

(C) Another example of conflicting prior probabilities. Here, node C changed its distribution significantly.

## PubMed Commons