implement callbacks to make problem modification pretty

[stelmo]

Problem modification needs to be streamlined

change constraints, solver options, objectives etc. in a nice, extendable way

[exaexa]

This depends on the previous PR right? can we mark the target branch accordingly? (so that the changes aren't mixed)

[stelmo]

This depends on the previous PR right? can we mark the target branch accordingly? (so that the changes aren't mixed)

yes, done!

[exaexa]

@stelmo I rebased this to master (which now also means that this resides correctly on top of #51). Before you start working on anything, please do git fetch -a; git checkout mo-callbacks-fba; git rebase --hard origin/mo-callbacks-fba. If you have any uncommited work, save it before using git stash.

[stelmo]

Despite you telling me exactly what to do, I think this is going to push my git skills to the next level 😁

[stelmo]

So I made many changes in commit cb91565:

1. CobraModel is now StandardModel
2. I implemented callbacks on all functions using StandardModel
3. Doing (2) caused me to simplify A LOT of code
4. I also created a new file called "utilities.jl" in src/analysis that should house functions that analyse solutions from functions in src/analysis, e.g. flux_balance_analysis
5. I changed the names of all my functions to fit in with your way of naming them e.g. fba -> flux_balance_analysis
6. I also made new function that map solutions to vectors and dicts like you guys do e.g. flux_balance_analysis_vec
7. I changed hit_and_run slightly: now you feed it a JuMP model as opposed to a "normal" model. Basically, I thought it makes more sense to do it this way since the user will almost always be interested in sampling from a solution space AFTER having run fba, this way they can just feed in the output of e.g. flux_balance_analysis and roll from there.
8. Functions that dispatch on StandardModel now return nothing if things went wrong during solving (like your functions)

[stelmo]

The next step is to get pretty macros working, currently I am trying to get this to work:

@flux_balance_analysis_vec model Tulip.Optimizer begin
    modify_objective(biomass)
    modify_constraint(glucose, -8.0, -8.0)
end

Unfortunately, I got stuck here trying to get the modification macro working :(

[stelmo]

The next step is to get pretty macros working, currently I am trying to get this to work:

@flux_balance_analysis_vec model Tulip.Optimizer begin
    modify_objective(biomass)
    modify_constraint(glucose, -8.0, -8.0)
    knockout(gene1)
    ...
end

Unfortunately, I got stuck here trying to get the modification macro working :( Note I have not implemented knockout(...), but the other modification functions are done.

I got this working! However, I stopped adding stuff to this PR since it was getting out of hand. Once it's merged I will make a new one and add more cool macros. Then we will be well on our way to making something amazing even more amazing :)

[stelmo]

Please see my replies

[laurentheirendt]

nice work 🎉