Add --binning, --zoomify and --balancing arguments to pipeline

[js2264]

• --binning INT: when --mat_fmt cool, a binned cool matrix can also be generated from pairs file;
• --zoomify: if added (and --binning INT is present), the binned matrix will be zoomified at binning * [1, 2, 5, 10, 20, 50, 100] and balanced
• --balancing STR if zoomify, the balancing options for the .mcool file can be specified (cooler default used by default)

So the suggested pipeline call could be something like:

hicstuff pipeline \
    --enzyme "DpnII,HinfI" \
    --matfmt cool  --binning=1000 --zoomify --balancing "--cis-only" \
    --genome <REF> \
    <R1> <R2>

I'm very tempted to change the default matfmt to "cool"...

[js2264]

Note: I added cooler to the requirements.txt... I hope this is ok?

[cmdoret]

Note: I added cooler to the requirements.txt... I hope this is ok?

It's currently listed in the extras_require (i.e. optional dependency) since at the time we did not use it as the main format.

I think it would make sense to make it a mandatory dependency, but then it should be removed from the extras_require.

Similarly, changing the default matfmt to cool sounds reasonable to me.

[cmdoret]

Awesome improvements!! I spotted a mistake and left a few suggestions :)

[cmdoret]

If we're adding a "binning" argument, should we disallow integers in enzyme?

[cmdoret]

Or maybe even remove enzyme, and only use it in cutsite. What do you think @ABignaud ?

[js2264]

Or maybe even remove enzyme, and only use it in cutsite. What do you think @ABignaud ?

We most likely should keep enzyme, AFAIU the filtering depends on it.

If we're adding a "binning" argument, should we disallow integers in enzyme?

Also, I believe enzyme=100 is used for mnase/dnaseC, no? So probably best to still allow integers

[js2264]

Ok we'll keep allowing integers for enzyme for now

[js2264]

@cmdoret thanks for your review, I've replied to your comments. Let's wait for @ABignaud input on enzyme