Skip to content

Changelog

All notable changes to HDDFlyzer will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

0.1.5 - 2026-06-15

Added

  • Added preparation for publishing HDDFlyzer to PyPI.
  • Added the manual publish-to-pypi.yml workflow using PyPI Trusted Publishing.

Changed

  • Updated project metadata to 0.1.5.
  • Updated environment.yml to install hddflyzer==0.1.5.
  • Updated CI to validate Python 3.11 and 3.12 with twine check.
  • Prepared the release workflow for a single build-and-publish path after the release tag and manual workflow dispatch are in place.

Validation

  • Validated the package on Python 3.11 and 3.12 before publication.
  • Validated wheel and sdist builds, twine check, and wheel-based smoke installation.
  • PyPI publication is prepared for this cycle but must still be triggered through the manual trusted-publishing workflow.

0.1.4 - 2026-06-13

Added

  • Added a GitHub Actions documentation workflow for MkDocs / GitHub Pages validation.

Changed

  • Updated the internal HDDFlyzer version to 0.1.4.
  • Separated documentation dependencies through the [docs] optional dependency group.
  • Added a preventive MkDocs version bound with mkdocs>=1.6,<2.0.

Documentation

  • Validated the documentation site with mkdocs build --strict.
  • Prepared the v0.1.4 cycle for documentation web / GitHub Pages readiness without publishing PyPI in this cycle.

0.1.3 - 2026-06-11

Added

  • Added minimal GitHub Actions CI for HDDFlyzer with local validation steps for tests, import smoke, CLI smoke, and package build.

Changed

  • Updated the internal HDDFlyzer version to 0.1.3.

0.1.2 - 2026-06-10

Added

  • Added global CLI version reporting with hddflyzer --version.
  • Added the short CLI version flag hddflyzer -V.

Changed

  • Updated the internal HDDFlyzer version to 0.1.2.

Packaging

  • Validated local editable installation with python -m pip install -e ..
  • Validated local wheel and sdist builds with python -m build.
  • Confirmed that hddflyzer/config/descriptor_config.json is included in both the wheel and sdist artifacts.
  • Kept PyPI publication deferred to a later release cycle.

0.1.1 - 2026-06-10

Changed

  • Reworked the project documentation site around clearer page identities: Usage now focuses on running HDDFlyzer and understanding generated outputs, while API Overview and Reference separates guided Python examples from the auto-generated API reference.
  • Refined the Home page narrative to better describe the motivation for HDDFlyzer, the dataset-first result model, descriptor-space provenance, and representative molecular features.
  • Updated the documentation visual identity with a teal/deep-petroleum palette, warmer admonitions, a single-color header, balanced dark-mode contrast, and more consistent footer spacing.
  • Improved the Changelog page rendering and presentation through MkDocs snippet support and focused changelog styling.

Added

  • Added Home-page examples for descriptor-space outputs using documentation assets from docs/assets/.
  • Added footer social links for the HDDFlyzer GitHub repository and the NanoBiostructures Research Group website.
  • Added a pre-stable documentation notice that states HDDFlyzer is currently in Alpha-stage development.

Fixed

  • Fixed the documentation Changelog page so it renders entries from CHANGELOG.md in the project documentation site.
  • Corrected the API navigation label from API Overview to API Reference, then refined the page title to API Overview and Reference to match its combined guide/reference role.

0.1.0 - 2026-06-08

Added

  • Added a CLI-first workflow for local cheminformatics analysis with hddflyzer as the user-facing command-line entry point.
  • Added a dataset-first results layout under results/<tag>/ so each molecular collection keeps its generated registry, annotations, chemistry outputs, dimensionality-reduction outputs, figures, metadata, and summaries together.
  • Added molecule-registry preparation from local input collections.
  • Added natural-product class annotation outputs.
  • Added Tanimoto similarity workflows, including similarity matrix output, molecule ID alignment files, sampling outputs, and metadata.
  • Added molecular descriptor feature generation, feature curation, correlation analysis, and correlation-based pruning outputs.
  • Added dimensionality-reduction workflows for PCA, t-SNE, and UMAP across descriptor, Tanimoto, and pruned-feature spaces where supported.
  • Added visualization commands for natural-product classes, similarity outputs, descriptor correlations, PCA, t-SNE, and UMAP projections.
  • Added manifest.json and workflow_summary.md as run-level indexes for generated outputs, operation metadata, current outputs, output categories, and workflow-contract reconstruction.
  • Added the programmatic pipeline contracts run_pipeline(), run_workflow(), and execute_workflow().
  • Added WorkflowExecution for programmatic workflow execution summaries, including stage results, failure tracking, and optional reconstructed run access.
  • Added the results API with WorkflowRun, ResultArtifact, LoadedArtifact, load_workflow_run(), and load_artifact().
  • Added semantic artifact classification for outputs such as molecule registries, descriptor tables, Tanimoto matrices, projection coordinates, figures, metadata files, and workflow summaries.
  • Added artifact selection and loading helpers through run.artifacts(...), run.artifact(...), and run.load_artifact(...).
  • Added the optional required artifact filter to disambiguate matching outputs by path fragment.
  • Added visualization input resolution through VizInputs and resolve_viz_inputs().
  • Added support for using reconstructed result objects as visualization inputs, including plot_hddf_scatters(LoadedArtifact) for loaded descriptor-table artifacts.
  • Added scientific result views over existing outputs: DescriptorSpace, SimilaritySpace, and ProjectionSpace.
  • Added WorkflowRun scientific accessors: run.descriptor_space(...), run.similarity_space(...), and run.projection_space(...).
  • Added molecular-identity utilities for comparing and aligning scientific spaces, including shared_molecule_ids(...), has_aligned_molecule_ids(...), and align_spaces(...).
  • Added scientific analysis helpers over reconstructed artifacts, including: similarity_projection_correlation(...), similarity_projection_neighbor_overlap(...), descriptor_projection_correlations(...), projection_neighborhood_preservation(...), and compare_descriptor_groups(...).

Changed

  • Consolidated the CLI around the programmatic workflow engine so hddflyzer pipeline run uses execute_workflow() while preserving existing exit-code behavior.
  • Preserved distinct execution contracts:
  • run_pipeline(...) returns list[StageResult];
  • run_workflow(...) returns a reconstructed WorkflowRun;
  • execute_workflow(...) returns WorkflowExecution.
  • Extended the run manifest with a workflow-contract block that describes the canonical registry -> chem -> dimred -> viz -> metadata/results workflow.
  • Added grouped output categories while preserving the existing flat current_outputs list for compatibility.
  • Refined the README to present HDDFlyzer as a local, traceable molecular descriptor-space workflow tool rather than a collection of unrelated scripts.
  • Clarified that “HDDFlyzer” is a project name/brand: HDD refers to high-dimensional descriptors, while Flyzer is a stylized name for exploring and analyzing molecular descriptor spaces.

Security

  • Blocked pickle loading by default in public artifact and dataframe loaders.
  • Required explicit allow_pickle=True for trusted local pickle files.
  • Added run-tag validation to reject empty tags, absolute paths, path traversal, and path separators.
  • Added containment checks so reconstructed artifacts must resolve inside their corresponding results/<tag>/ directory.
  • Added manifest-update checks that reject registered output files outside the expected run directory.
  • Made matrix loading use non-pickle NumPy loading by default.

Documentation

  • Documented the traceable results model based on results/<tag>/, manifest.json, workflow_summary.md, artifact semantics, and reconstructed runs.
  • Documented scientific result views and analysis helpers as operations over already generated artifacts.
  • Documented that the scientific API does not recalculate descriptors, similarities, or projections, and does not perform automatic clustering, enrichment, or chemical interpretation.
  • Documented local safety expectations for pickle files, path validation, and untrusted manifests.

Tests

  • Added and expanded tests for CLI routing, configuration, IO, manifest writing, pipeline execution contracts, results reconstruction, artifact loading, visualization inputs, public API exports, scientific spaces, alignment, and scientific metrics.
  • Added tests for defensive path handling, tag validation, artifact containment, and pickle-blocking behavior.

Notes

  • HDDFlyzer is not yet a complete drug-discovery platform, docking workflow, automatic chemical-interpretation system, or automatic enrichment/clustering tool.
  • Current workflows focus on local molecular descriptor spaces, similarity matrices, dimensionality reduction, visualization, result reconstruction, and lightweight scientific analysis over existing outputs.
  • Future conformer-generation and 3D descriptor workflows are planned but are not part of the current workflow contract.