need a little cleanup and docs on variants

[1fish2]

• Which of the variant types are useful and working? Should the others be deleted?
• EndoKcatFullRNA.py and rnaDegRateShuffleParams.py aren't listed in nameToFunctionMapping and nameToNumIndicesMapping so the code won't find them. Should these be added to the list? Deleted?
• The keepers could use docstrings.
• What are some variant types and variant indexes recommended for people to try?
• Which variants implement cell variations that make sense within the multi-scale environment simulation? I'd expect that geneKnockout applies while nutrientTimeSeriesDownshift and anaerobic don't.

EndoKcatFullRNA  # inaccessible
allShuffleParams
anaerobic
catalystShuffleParams
condition
gene_knockout
kineticCatalystShuffleParams
kineticTargetShuffleParams
kineticsConstraints
meneParams
metabolism_kinetic_objective_weight
metabolism_target_range
monomerDegRateShuffleParams
nutrientTimeSeries
nutrient_time_series_downshift
rnaDegRateShuffleParams  # inaccessible
scaling_factor
starvation_variant
tf_activity
time_step
transcriptionInitiationShuffleParams
translationEfficienciesShuffleParams
wildtype

[tahorst]

Most of these have fallen out of step with the actual code. I think we can remove a fair number and should probably think about having a quick jenkins build to test any we want to keep.

It would also be useful to standardize the notation to underscore_function_names. There are a few files with underscores but the functions are camel case so to properly run the variant, you would need to set VARIANT=tfActivity for the python fw_queue.py call even though the variant file is named tf_activity.py. I think it creates some confusion with the mixed standard and moving forward we should use underscore based on our style guidelines.

Variants to remove:

• anaerobic - covered by condition variant 1 which is tested in the daily build
• EndoKcatFullRNA - probably can remove (sim_data.KcatEndoRNaseFullRNA isn't used anywhere so at the very least would need to be updated)
• growth_rate - will fail passing doubling_time argument to fitSimData_1
• kineticsConstraints - sim_data.process.metabolism.useAllConstraints is not used and running metabolism_kinetic_objective_weight variant 0 would have the intended effect
• metabolism_target_range - sim_data.constants.metabolismTargetRangeConstant isn't used and we don't use the metabolism objective it is trying to test
• nutrient_time_series_downshift - this should be handled by nutrientTimeSeries
• scaling_factor - sim_data.scaling_factor isn't used, not sure what it was trying to test
• starvation_variant - attributes it sets don't get used
• time_step - sim_data.timeStepSec doesn't get used

Variants to update:

• gene_knockout - will fail on sim_data.process.transcription.rnaData["synthProb"] since it's no longer part of rnaData and expression is handled differently now but the possibility of doing knockouts is probably still useful
• nutrientTimeSeries - it doesn't update sim_data.condition like the condition variant does so a cell will be initialized to basal conditions even if it starts in different nutrients

Additionally, all of the *ShuffleParams variants could probably be removed. We really only added it for a figure in the old version of the paper. Removing this would also require cleaning up some additional code in process files and analysis files.

meneParams and metabolism_kinetic_objective_weight were used in the paper and might have limited utility moving forward (metabolism_kinetic_objective_weight could probably still be useful for future debugging). tf_activity is also probably only useful for debugging.

That means that the only really useful variants at this point are wildtype and condition which are both tested with Jenkins. I agree that we can add some docstrings to the ones we keep and can describe the useful indices and the intentions of the variant in those docstrings.