Toy example on reachability, fixed point, and mutant

This is a simple example of BoNesis usage for Boolean network synthesis from properties combining reachability, fixed points, and mutants.

import bonesis

We take as domain any possible Boolean network between 3 variables, and define 2 observations: one (complete) for the initial configuration, and one (incomplete) for the final one:

# use complete graph as domain
dom = bonesis.InfluenceGraph.complete(["A","B","C"])
# named observations
data = {
    "init": {"A": 0, "B": 0, "C": 0},
    "marker": {"A": 1, "C": 0}
}

We declare that, in the wild-type, there is a configuration matching with the final observation ("marker") which is a fixed point reachable from the initial configuration. Moreover, in the mutant where C is KO, the final configuration is no longer reachable.

bo = bonesis.BoNesis(dom, data)
x = ~bo.obs("init")
y = ~bo.obs("marker")
x >= bo.fixed(y)  # y is a fixed point and can be reached from x
with bo.mutant({"C": 0}):
    x / y

Existence of a solution can be verified as follows.

bo.is_satisfiable()

True

Then, we enumerate 3 solutions, and display the first one. Solution objects are mpbn.MPBooleanNetwork objects (https://mpbn.readthedocs.io).

solutions = list(bo.boolean_networks(limit=3))
solutions[0] # show one solution

A <- A|(!B&C)
B <- !B|(A&!C)
C <- !B

Computation of (most permissive) attractors can then be performed using mpbn as follows.

# list attractors
import pandas as pd
pd.DataFrame(solutions[0].attractors())

	A	B	C
0	1	1	0

The Boolean network can be exported to the standard BoolNet textual format, which is supported by many tools for analyzing Boolean networks.

solutions[0].save("solution.bnet")