v0.2.2

New Features:

1. support cryoem mode, using electron scattering factor
2. support mmcif file as PDB input, cif file as reflection input
3. support to skip the solvent calculation, even in batch calculation
4. a new get_rfactor method

Bug Fix:

1. Handle edge case where no space group in PDB file
2. More tolerance in cell isomorphism check
3. Take care of systematic absence indices
4. Write PDB with TER records

Full Changelog: v0.2.1...v0.2.2

v0.2.1

1. Change the implementation in get_scales to make the computation graph more robust

2. Tackle the edge cases where no freeflag or zero Fo values are detected.

3. Include sequence property in PDBparser

4. Bug Fixes

v0.2.0

A major version bump with:

1. Better typing hints
2. Support more flexible constructor, now users can input path strings to pdb/mtz files, or existing PDBparser or rs.Dataset instance
3. Use getter and setter to make the molecule-related properties more robust
4. A new savePDB method in SFcalculator, support to directly save refined model out
5. Bug fixes

v0.1.10

1. Remove torch from pip requirements in setup.py, in case it will destroy some delicate pytorch envs
2. Fix the multi device bug

v0.1.9

1. Unify the device setting of SFCalcualtor
2. A few engineering tricks to be memory efficient in batched Fprotein and Patternson calculation
3. Update default hyperparameters in calc_fsolvent based on statistics
4. deprecate 3.7 support
5. A new function to fetch PDB with static urllib
6. bug fixes

v0.1.8

Update API

v0.1.7

1. update methods api, to use lowercase letters
2. support logarithmic algorithm to assign bin labels,
3. change to binwise scaling, and use lbfgs to update
4. add a summarize method
5. fix the bugs related to the DWF_aniso calculation (finally)

v0.1.6

1. support anomalous scattering
2. use tiny datasets in unit tests
3. support patterson calculation

v0.1.5

First automatic release, with decent unittests now