Protein inheritance

[Yuri05]

At the moment only complete building blocks (and for simulations: simulation parameter values) can be inherited from another project within a qualification plan:

However for Individuals/Populations it might be required to onherit not the whole building block but only the proteins.
For this, qualification plan could be extended with the Protein Inheritance feature, e.g. something like this:

Then Protein (expressions) of a building block would be replaced by the protein expressions of the parent.

Question / Topics to discuss:

1. Should proteins always be replaced or should it be possible also to merge? E.g. if the parent-BB contains proteins E1, E2 and the child-BB contains proteins E3, E4: would the result always contain only E1, E2 or E1, E2, E3, E4?

2. Should protein inheritance be defined only for individuals? Or also for populations?

3. For proteins previously not available in a simulation: how should we proceed with the simulation mapping Compound-Process <=> Individual Protein ? In my eyes is this the most critical point of the whole discussion

4. If previously available proteins are deleted by the inheritance: how should we proceed if e.g. curves related to a deleted protein were referenced in the project (e.g. plotted/used in a PI/...)?

[Yuri05]

1. Should proteins always be replaced or should it be possible also to merge? E.g. if the parent-BB contains proteins E1, E2 and the child-BB contains proteins E3, E4: would the result always contain only E1, E2 or E1, E2, E3, E4?

I would vote for "replace only". No merge.

[Yuri05]

3. For proteins previously not available in a simulation: how should we proceed with the simulation mapping Compound-Process <=> Individual Protein ? In my eyes is this the most critical point of the whole discussion

Many possible solutions here and a lot of space for side effects.
If Individual (or population) in the qualification plan inherits its proteins, then all simulations where this individual/population is used should become an info "add default mappings for this particular individual/population" or something in this direction. Thus:

• if a previously available protein was removed from the individual/population due to inheritance: its mappings must be removed from the simulation configuration
• if a new protein was added: the default mappings for this protein would be used in the simulation (which is pretty error prone and migth be NOT what the user has in mind! e.g. right now the default mapping takes the compound process which was created first)
  ➡️

Thus either something like default flag must be added to compound process definition.
Or the mapping must be defined in the qualification plan (which I would NOT do for sure!)

Even if something like "default process" would be defined in a compound, there are still situations where the default mapping would not create everything what the user wants (e.g. in case where several metabolites are created by the same process; some DDI combinations, etc.)

To be honest, the whole procedure is so complicated and error-prone...

[PavelBal]

1. I definitely vote against deleting proteins!

I do not fully understand the problem with adding new proteins. If protein A is added by the inheritance to a simulation, then the configuration of such a simulation does not has a mapping of any compound to the protein, right?

[Yuri05]

If protein A is added by the inheritance to a simulation, then the configuration of such a simulation does not has a mapping of any compound to the protein, right?

Right. If we keep it like this (new proteins are just added to the building block but not used in simulations) - much easier to implement. However: what is then the benefit of the feature? Just the possibility to update the properties (relExp etc.) of previously existing proteins?

I definitely vote against deleting proteins!

This would solve some of the problems for sure :)