v1.0.1

Small changes and additions to kissim:

• Dependency fixes (pin ipywidgets, add tqdm, remove biopandas)
• CI workflow updates
• Add residue subset by Martin et al.

What's Changed

• Update test_env.yaml by @dominiquesydow in #104
• Update CI (OS and Python versions) by @dominiquesydow in #107
• Add residue subset by Martin et al. by @dominiquesydow in #108
• Fix docs CI by @dominiquesydow in #109
• Update CI schedule to Mondays 3 am (UTC) by @dominiquesydow in #111
• Pin ipywidgets to <8 by @dominiquesydow in #113

Full Changelog: v1.0.0...v1.1.0

v1.0.0

This is the first official kissim release!

What's Changed

• Compress fingerprint/feature distances output files (csv.bz2) by @dominiquesydow in #95
• Fix bug in viewer module by @dominiquesydow in #96
• Add Phylo tree to docs by @dominiquesydow in #97
• Update kissim.viewer (NGL GUI optional + absolute FP diff) by @dominiquesydow in #98
• Pin matplotlib to <3.5 by @dominiquesydow in #101
• Fix kissim.viewer discrete colorbar by @dominiquesydow in #100
• Add conda-forge channel to installation docs by @dominiquesydow in #103

Full Changelog: v0.4.1...v1.0.0

v0.4.1

Minor changes

Package

• Specify logger levels for kissim (INFO) and opencadd (WARN) #90
• Fix future warning (df.var(level=1) > df.groupby(level=1).var()) in kissim.viewer.kinase.KinaseViewer._std #94

Environment

• Use opencadd conda package in environment (instead of pip install of GH repo) #91
• Use matplotlib-base instead of matplotlib in environment #93

v0.4.0

New

• Min-max normalization for continuous values more fine-grained + CLI (kissim normalize)
  • Distances: Min-max per subpocket and residue position
  • Moments: Min-max per moments and subpocket
• Add residue subsets for fingerprinting + CLI (kissim subset)

Fixed bugs

• Sort kinase pairs alphabetically before groupby (important to have unique pair combinations for structure to kinase mapping)
• Add exceptions for errors detected with updated KLIFS dataset

v0.3.0

This release contains a lot of big changes to the code base and new features. (In the future cut releases earlier!)

New

Package

• Replace angles and ratios in SideChainOrientationFeature and SolventExposureFeature with categories (1, 2, 3) PR
• Add viewer module with StructureViewer, StructurePairViewer, KinaseViewer PR and PR
• Use opencadd.databases.klifs to fetch structural information from KLIFS (local or remote) PR
• Use opencadd.structure.pocket to calculate subpocket centers; update anchor residues so that less fingerprints contain NaN values because of missing residues PR
• Add CLI with sub-commands (kissim encode; kissim compare; kissim outliers; ...) including a structure outlier removal option PR and PR
• @AJK-dev: Add kinase matrix clustering and tree generation (kissim.comparison.tree module) PR
• Save FeatureDistancesGenerator and FingerprintDistanceGenerator data as CSV files and FingerprintGenerator data as JSON file PR
• Add fingeprint pair bit coverage minimum option PR

Repo structure and development workflow

• Rename repo from kinsim_structure to kissim PR
• Rename master to main branch
• Add unit tests and CI for Linux/MacOS/Windows
• Add tutorials for each module; set up RTD documentation
• Add CodeCov

Changes

Package

• Change side chain orientation calculation PR
  • Before: Calculate angle between pocket centroid, residue CA atom and residue's side chain centroid
  • Now: Calculate angle between pocket centroid, residue CA atom and residue's side chain representative (furthest atom from CA) and map angles into categories
• Use NGLview instead of PyMol for visualization
• Remove MOL2-to-PDB conversion; use KLIFS started to offer PDB files directly
  • Add PocketBioPython and PocketDataFrame as data classes for the encoding step
  • Add KlifsToKissimData (streamline KLIFS data into KiSSim's required data format)
• Restructure code base (split large files into a module with multiple submodules)
• Rename similarity > comparison module PR
• Rename ExposureFeature > SolventExposureFeature as part of this PR
• Fingerprints are not normalized by default any more; instead distances are calculated per feature type (thus dimensions match) and can be weighted user-defined to produce the final fingerprint distance PR
• Move structure/kinase matrix generation in FingerprintDistanceGenerator class method to kissim.comparison.matrix module PR

Repo structure and development workflow

• Update to cookiecutter-cms 1.5
• Move all analysis notebooks to https://github.com/volkamerlab/kissim_app
• Move KLIFS dataset preparations to https://github.com/volkamerlab/kissim_app

Bug fixes

• Fix SiteAlign feature implementation for THR's charge (was positively charged but must be neutral) PR