1.0.0

[orbeckst]

The 1.0.0 release of alchemlyb comes with a stable API and multiple enhancements for postprocessing and workflows.

Importantly, the 1.x release of alchemlyb only supports pymbar >=3.0.5, <4. We plan to switch to pymbar >= 4 soon (see #205 , #207) and then release a 2.0.0 of alchemlyb. We will not backport fixes from 2.x to 1.x or maintain a 1.x for the long term (unless volunteers from the community offer to take on maintenance).

Supported platforms

We follow NEP29 for supported Python/numpy versions. Currently supported:

• Python 3.8 – 3.10 (continuously tested), 3.11 (manually tested)
• Linux, macOS, Windows
• pymbar >= 3.0.5, < 4

Packages are available from PyPi and conda-forge.

Highlights

• stable API
• complete absolute binding free energy (ABFE) workflow (Implement the workflow pipeline #114, ABFE workflow will not read dHdl file if no TI estimator is choosen #231) in the new workflow module (Implement the base workflow class #188)
• more robust AMBER parser (Amber parser incorrectly set the time when reading dHdL/ u_nk values #221, AMBER: temperature reading in parsers and tests.  #225, AMBER: why are we reading AMBER output averages (and not using them)? #226, AMBER: inconsistent use of raise Exception / logger.exception #227, AMBER: regarding #FIXME and #NOTE about reading integers/float while parsing files #229, AMBER: are we too permissive in reading the files? (and also we are doing a double-reading? #238)
• new fractional equilibration time convergence analysis ( $R_c$ and $A_c$ from Fan et al 2021)
• common extract(file, temperature) function to read both dHdl and u_nk data in one pass if possible (Amber parser should be improved #222)

For all changes in this release see CHANGES.

Contributors

• @xiki-tempula
• @DrDomenicoMarson
• @orbeckst

---

This discussion was created from the release 1.0.0.