I tried to create some mutations (using model.mutate or maboss.copy_and_mutate) and then wanted to copy the model to apply different changes: the mutations are missing from the copy.

Digging into the code, the network contains two copies of each node: one in network.nodeList and the other as dict value. The mutation is only changing the dict value, while the copy restarts from the nodeList.

I'm working on a patch to stop using a separate list of nodes and only rely on the inner dict (we can still use the list of names to get the order).

The constructor for a Simulation uses a default list of parameters as starting point. As the same dict is copied in all objects, we always end up with the same parameter set.

It can be solved by adding a call to copy() to the line self.param = _default_parameter_list

The current parser works well with the bnd/cfg files generated by GINsim, but fails on many handwritten files. Supporting all of MaBoSS syntax is a complex task, beyond the scope of the current parser, but we can handle a few more cases with minimal work.

• The "node" keyword should ignore case
• node.istate should work with "true" and "false" keywords
• variable assignments can be more complex than a simple value (1/5 instead of 0.2 is common)

The first two are easy to fix (pull request is following). For the variable assignment, it may be simpler to just store the value as string without trying to interpret it (it can depend on another variable): are there some parts of the code where we need to work with these values?

My last problem is that the current parser will ignore silently everything that appears after an unrecognized line: it should try to parse the remaining and show a warning message, maybe keep the failed lines as string to be able to write them in print_cfg() anyway.