TO_DO

==============
==  TO DO:  ==
==============

March 6:
========

 BIG GOAL:
 	- Improved summary of red-check data with pictures for EVERY sample,
 	  cross-referenced with Matt's notes, and methods tuned for that data.
 	- Summary of bicycle model compared to data.

 - Create data processing tools
 	- Write functions for ease of visualization
 	- Implement objective evaluation functions
 	- Write functions for ease of visualization of data with objectives results
 	  on top.
 	- Write functions that make it very easy to load data.
 		- Provide a template for how to do simple things:
 			- Load a date
 			- Load a single file
 			- Load several dates
 			- Load all of specific protocol from given dates
 	- Write tool to cross-reference with red_checks_data.csv
 		- Report date of data, ZT, genotype, parameters that Flourakis measured

 - Write functions to generate simulation protocol structure.
 	- Based on data
 	- By hand?
 	- What should this structure be?

 - Port over abf_util to MOO_CNM (should just be copy/paste)

 - Create summary figures of each objective over time
 	- FR, MemPot, SpikeAmp, RI, tau, C
 		- (Fitted) scatter plots
 		- Box plots

 - Show representatives for each time bin

 - Implement Bicycle model... (do this with Individual class if possible!)
 	- Show its features under different standard protocols
 	- Write simulation functions
 	- Write Individual visualization functions

 - In ephys_objs 'dt' will be grabbed either from infoDict['data'] or
   infoDict['simulation']


Overall:
========

 - Create runfile which does:
 	- Load and check runfile (info.json)
 	- Process data, create data features structure and simulation protocols
 	- Set up parallel structure
 	- Run the whole shebang (distribute evaluation of individuals)

 - Create functional Population class that can
 	- Mutate
 	- Crossover
 	- Evaluate (non-parallel)
 	- Select Best Individuals

 - Create functional Individual class that can
 	- Load model function and parameters
 	- Mutate
 	- Crossover with another Ind
 	- Evaluate itself
 	- Simulate itself??
 	- Visualize simulations?

 - Create utilities for
 	- Protocol dictionaries
 	- Data I/O
 	- Working with ABF files
 	- General data checking
 	- Processing info

 - Create objectives functions
 	- Functions to parse dataFeat
 	- Functions to create dataFeat
 	- Functions to measure:
 		- ISI, MemPot, SpikeAmp, RI, tau, C
 		- Reliably and automatically

 - Create models
 	- Functions for simulating models according to protocols
 	- Functions for inferring protocols from data
 	- Functions for visualizing models
 		- Decide how to store trajectories...
 		- Probably good to have tool to easily create new trajectories.

 - Create functional Hall of Fame class
 	- Update!
 		- Using stereographic projection, medians need to be updated!?
 	- Is there a way to parallelize this step?

 - Think about:
 	- Protocol structures
 	- Data features structures
 	- Generating fake data from model, test procedures
 	- Post-run processing
 		- How to visualize Pareto Front
 		- Visualize convergence!

 - runfile_util.check_defaults should only replace *necessary* or *convenient*
   parameters, not all...
   	- See infoDict['data'] when input disagrees with default.
   		- Or any case where we just want to check that there is *any* key, not 
   		  a specific one.

 - runfile_util checking objectives should check and set keywords for all
   pertinent ephys_objs keywords.
   	- "Spike":{}  Should be checked to have all keywords that go into
   	  getSpikeIdx, for example.
 - Check that infoDict['objs']['spikes']['maxRate'] == ['ISI']['maxRate']
 - Eliminate 'kwds' field from objectives.  Just use **iDict[objs][obj].

 - Verify that Data Numbers Match Flourakis Reported Data

In Progress:
============
 - Runfile checking
 - Individual
 - Population
 - Model Implementation

Finished:
=========
 - Sketched generic structure
 - Ported abf_util
 - Fixed / rewrote utility library
 - Wrote DataIO_util