Instanovo

What's Changed
Code updates for publication release
* fix: add sample data by BioGeek in https://github.com/instadeepai/InstaNovo/pull/81
* docs: update readme for publication by KevinEloff BioGeek rcatzel in https://github.com/instadeepai/InstaNovo/pull/82
* fix: add diffusion device option, set automatically on load by KevinEloff in https://github.com/instadeepai/InstaNovo/pull/83
* fix: use separate diffusion sdf instance, minor release bump by rcatzel in https://github.com/instadeepai/InstaNovo/pull/84


**Full Changelog**: https://github.com/instadeepai/InstaNovo/compare/1.1.0...1.1.1

What's Changed
* Updated to use [Typer](https://typer.tiangolo.com/) BioGeek
* Typer includes options to predict with InstaNovo and refine with InstaNovo+ in one command
* Updated dependency management to use [UV](https://docs.astral.sh/uv/) BioGeek
* Re-added updated diffusion code rcatzel
* A preliminary checkpoint has been added
* Adds support for UNIMOD ProForma format in the residue class (backwards compatible with original format)
* Adds automatic UNIMOD conversion of old model checkpoints during loading by default
* Adds support for specifying a list of globs for SpectrumDataFrame
* Removed (+25.98) N-terminal modification from training, and added to suppressed list by default
* Fixed bug where tempdir would not be initialised if all input data is in `.parquet` format, causing a crash when performing a preshuffle.
* Updated predict script to always calculate `delta_mass_ppm` even in _de novo_ mode.
* Updated README.md
* Added usage for UV and Typer
* Added link to Acknowledgements
* Added **Natively Supported Modifications** table with unimod ID
* Added **Output description** table
* Updated **Using your own datasets** table to remove the `modified_sequence` column

New checkpoints
* Trained a new checkpoint with corrected cysteine modification and added MassiveKB data: **instanovo-v1.1.0**
* Preliminary InstaNovo+ checkpoint trained on the same data: **instanovoplus-v1.1.0-alpha**

Notes:
* A bug with Implicit Cysteine modification has been corrected in the AC-PT. This bug affects the previous `instanovo_extended.ckpt` checkpoint from the [1.0.0 release](https://github.com/instadeepai/InstaNovo/releases/tag/1.0.0). This bug caused significant performance issues on downstream datasets.
* The diffusion checkpoint is an alpha release with an updated checkpoint coming soon.

* fix: Resolve "AttributeError: 'SpectrumDataFrame' object has no attribute 'df'"

* feat: update notebooks to v1.0.0

* feat: Automatic model download and improve residues
Co-Authored-By: Kevin Eloff <k.eloffinstadeep.com>

* feat: update tests for v1.0.1 release
Co-Authored-By: Rachel Catzel <r.catzelinstadeep.com>

* feat: update packages

New Contributors
* dependabot made their first contribution in https://github.com/instadeepai/InstaNovo/pull/55

**Full Changelog**: https://github.com/instadeepai/InstaNovo/compare/1.0.0...1.0.1

Improved code utility and data validation
- Check labels match precursor
- Check for data leakage
- Verify residue vocabulary
- Added better residue support
- Fine-tuning trainer automatically updates model weights with new sizes
- Added Flash attention, torch.compile(), AMP (fp16)
- Added improved fast greedy search
- Improved test coverage

Added Spectrum Data Handler
- Supports lazy loading with asynchronous prefetching
- Filtering and sampling performed non-destructively (by updating the row filter)
- Two-fold shuffling strategy for training ensures optimal load times

Extended model checkpoint released. Trained on 32M spectra with additional PTMs:
- AC-PT
- Additional PRIDE dataset
- Additional phosphorylation dataset

**Full Changelog**: https://github.com/instadeepai/InstaNovo/compare/0.1.6...0.1.7

- Add data conversion scripts
- Add InstaNovo+ to starter notebook
- Add functionality to return all beams (`return_all_beams=True` argument in `decoder.decode`) 12