Add option to not save generated data

[swkeemink]

When defining an experiment it is now possible to define the format in which data is saved with the data_format option. At the moment this can only be 'npy' or None. If it is 'npy', the behavior will be as before. If it is set to None, no data will be saved. This is useful for users that don't need to store any of the data directly on the hard drive, and just want to define how and where to use the data in their own workflows.

I have also laid the basic groundwork for other formats than npy, and am working on saving everything in hdf5 files.

Note that there is currently a bug with our use of numpy.load, see #85.

[scottclowe]

If I understand correctly, you're intending to merge this as-is with the option of not saving added, and HDF5 support will be in a different PR?

[swkeemink]

Correct!

[scottclowe]

Better code style is to have the ): on its own line, so you don't get this diff where a comma and a new line are inserted later on (which happened here).

Suggested change

	lowmemory_mode=False, datahandler_custom=None):
	lowmemory_mode=False, datahandler_custom=None,
	data_format='npy'):
	lowmemory_mode=False, datahandler_custom=None,
	data_format='npy',
	):

[swkeemink]

Very good point, I will change this.

[scottclowe]

Shouldn't add comments about prospective future features to docstrings.

[scottclowe]

I think you need to specify the behaviour better. At the moment it isn't clear what happens if None is provided.

[swkeemink]

Will clarify this in the docstring.

[scottclowe]

Suggested change

	data_format : string or None, optional
	How FISSA generated data will be saved.
	Can be 'npy' (default) or None. In the future will also support 'hdf5'.
	data_format : 'npy' or None, optional
	The format in which data generated by FISSA will be saved.
	If this is `'npy'`, data will be saved in the numpy format using
	`numpy.save`. If this is `None`, the data will not be saved to
	a cache file and will have to be generated again if the same
	data is separated again. Default is `'npy'`.

[scottclowe]

Remove superfluous nesting.

Suggested change

	# try to load data from filename
	if not redo:
	try:
	nCell, raw, roi_polys = np.load(fname)
	print('Reloading previously prepared data...')
	except BaseException:
	redo = True
	# try to load data from filename
	if not redo:
	try:
	nCell, raw, roi_polys = np.load(fname)
	print('Reloading previously prepared data...')
	except BaseException:
	redo = True

[swkeemink]

Besides the above changes (which I generally approve of) we should also expand the tests to avoid reducing code coverage as it is now.

[codecov]

Codecov Report

Merging #86 into master will decrease coverage by 1.33%.
The diff coverage is 58.62%.

@@            Coverage Diff             @@
##           master      #86      +/-   ##
==========================================
- Coverage   79.27%   77.94%   -1.34%     
==========================================
  Files           9        9              
  Lines         666      680      +14     
  Branches      129      136       +7     
==========================================
+ Hits          528      530       +2     
- Misses         94       99       +5     
- Partials       44       51       +7

Impacted Files	Coverage Δ
fissa/core.py	85.52% <58.62%> (-4.67%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 50f148c...0a41db9. Read the comment docs.

[scottclowe]

Closed in favour of #129, which handles the ability to run fissa without saving to a cache more succinctly. @swkeemink please make a new PR for saving to HDF5 format.