Releases: PaccMann/paccmann_datasets
Releases · PaccMann/paccmann_datasets
v1.1.7
What's Changed
- feat: allow protein tokenizer to be pickled by @jannisborn in #166
-> Support for pytorch lightning ~ > 1.5
Full Changelog: v1.1.6...v1.1.7
v1.1.6
What's Changed
- Crawlers by @jannisborn in #165
- Fix a bug in the PubChem crawler that occurs when the database has multiple SMILES for a molecule
- Removing zinc crawler because it is not functional
- Restructure installation to support recent versions of numpy (2), torch (2), pip (20-24) and python (3.9-3.12)
Full Changelog: v1.1.5...v1.1.6
v1.1.5
v1.1.4
Full Changelog: v1.1.3...v1.1.4
Improved PubChem error handling
v1.1.2
What's Changed
- feat: bump selfies version (close #150) by @jannisborn in #162
Full Changelog: v1.1.1...v1.1.2
v1.1.1
Kinase sequence alignment data now available as pytoda.proteins.kinase_as_alignment: (245c9b5)
Full Changelog: v1.1.0...v1.1.1
v1.1.0
What's Changed
- Deprecate device handling by @YoelShoshan in #156
NOTE: This DOES impact backwards compatibility because now, whenever a device is passed, an exception will be raised saying that GPU support is not longer maintained. The reason for this is that having pytoda handling sending data to the GPU is significantly slower compared to sending the full batch (for details see: #155) - Novel protein augmentation by @jannisborn and @YoelShoshan in #160
Full Changelog: v1.0.2...v1.1.0