Add as_dataframe to EffectCollection

[arahuja]

This adds an as_dataframe to EffectCollection - happy to add or rename any columns.

Also, a few helper properties to Variant that I have been using:

• is_snv
• is_transition
• is_transversion

---

This change is

[iskandr]

Very slight nitpick: what do you think of to_dataframe instead of as_dataframe? It feels slightly more "pythonic" and seems to be the preferred method name for other projects: BigQuery, django-pandas, xarray.

[arahuja]

Ah yea, I like to_dataframe as well, but wanted to match the issue #128

[tavinathanson]

Issue #128?

[arahuja]

Yes, updated, I meant #128

[iskandr]

Twice?

[arahuja]

Woops, fixed

[coveralls]

Coverage increased (+0.3%) to 87.065% when pulling 922901d on effects-df into d46b2c5 on master.

[coveralls]

Coverage increased (+0.3%) to 87.065% when pulling 922901d on effects-df into d46b2c5 on master.

[coveralls]

Coverage increased (+0.2%) to 86.989% when pulling 922901d on effects-df into d46b2c5 on master.

[coveralls]

Coverage increased (+0.3%) to 87.065% when pulling 922901d on effects-df into d46b2c5 on master.

[arahuja]

@iskandr seem to be running into the problem you have on pyensembl, where the py 2.7 tests on Travis on failing with The command "nosetests test --with-coverage --cover-package=varcode && ./lint.sh" exited with 137.

[iskandr]

@arahuja Should be fixed now but have to wait for Travis to run. The problem was that I was decoding byte strings into ASCII, which does the sane thing under Python 3 but under Python 2 causes us to leave the byte-per-character str type to a fatter unicode representation which uses 2 (or 4!) times as much space per character. The dictionaries of transcript sequences are about ~200-300mb per release, so with 4x larger characters it's easy to imagine how we blow through Travis's 3gb limit.

[arahuja]

@iskandr Looks to be passing now

[arahuja]

@iskandr Good to merge?

[iskandr]

@arahuja Definitely

[timodonnell]

might make sense to be clear here about the coordinate system, e.g. call it inclusive_start. Just a suggestion, feel free to ignore

[arahuja]

Similarly, I'll make it match the varcode field names

[timodonnell]

these are the column names i've been using. Not necessary to match them but just fwiw https://github.com/hammerlab/varlens/blob/master/varlens/variants_util.py#L26

[coveralls]

Coverage increased (+0.2%) to 86.989% when pulling 1421b59 on effects-df into d46b2c5 on master.