Pointer to zenodo links (#185)

update __version__

Merge branch 'main' into jjimenezluna/release1

fix: don't require `ensurepip` (#120)

It seems that the system-installed python in Colab does not have
`ensurepip`. This PR removes the need to have this by using
`--without-pip` option. In addition, it moves `uv` from the colabfold
setup script to `pyproject.toml` which makes the dependency more
transparent.

Revert "chore: tie colabfold dir to the python executable (#104)" (#118)

Tying the colabfold directory to the python executable seems to have
been unnecessary in retrospect and caused other headaches. We initially
thought multiple python environments cannot share a patched colabfold,
but we believe this was false, because patched colabfold is a
self-contained python environment. Therefore, I suggest we revert it.

This reverts commit 159953e.

---------

Co-authored-by: José Jiménez-Luna <[email protected]>

fix: remove debug slice (#107)

PR removes
- missed debug code from #100
- pretty-print from typer to avoid printing all atomic positions when an
exception is raised

bump version to `0.1.9` (#106)

fix: more reliable MD relaxation & free MD (#100)

PR fixes
- issues resulting from proPka protonation by moving to openMM's native
protonation (less accurate, more reliable)
- a bug from using wrong units in MD integrator
- openMM pip installation did not run CUDA

PR introduces
- much more powerful equilibration protocol that can handle clashes
- unconstrained MD simulations as optional user choice


Some of the issues related to MD relaxation (e.g., #61) might have been
caused by using propKa to get protonation states. OpenMM has problems
reading those and in many cases does not recognise non-standard
protonation of residues, even if they have temples in the used force
field. To avoid the overhead of patching this, I suggest using openMM's
native hydrogen assignment. It might not yield accurate protonation
states of histidines etc, but this code is for relaxing samples only and
the approximation should be fine for this purpose.

---------

Co-authored-by: José Jiménez-Luna <[email protected]>

bump version

fix: MD relaxation with multiple samples (#81)

The MD relaxation step was failing when sampling from bioemu with
multiple samples due to an indexing mismatch.


![image](https://github.com/user-attachments/assets/2cd8871a-c2d5-4897-9f92-7b4be98cf25b)