Angle fix

[brookehus]

Development:

• [x] Implement feature / fix bug
• [x] Add documentation

• [x] Add tests

  Checks:

• [x] Run nosetests
• [x] Check pep8 compliance

Here is a fix for the angle stuff. The angles were done not by the normal convention (always subtracting the vertex point from the non vertex point) and weren't the same as mdtraj. The convention has been fixed.

[brookehus]

Since there's a lot of formatting edits all around, the major changes are:

base, offset = self.get_vectorize_inputs(angle_inds, data)
base *= -1

where the multiplication is new, and in

in get_angles in feature/geometry.py, and the corresponding change in test_backbone_angle_features in tests/test_geometry_features is:

ba = a - b
cb = c - b

where before it was ba = b - a. The changed convention always subtracts the vertex point b.

[nec4]

I think that this works well. Here is a comparison of GeometryStatistics (orange) and mdtraj.compute_angles (blue) before the changes:

[nec4]

and the same plot after the changes:

[nec4]

Tests also pass for me.

[nec4]

Other than one tiny doc suggestion, LGTM!

[nec4]

LGTM!