| name | app_integration |
| description | How the Streamlit app and the reusable mkt.databases.app package fit together (SequenceAlignment, StructureConfig/Visualizer, PropertyTables) and the schema → databases → app dependency chain. Use when editing the Streamlit app or the mkt.databases.app modules, or wiring visualization logic between them. |
App integration (mkt.databases.app ↔ Streamlit app)
Two pieces, clear split
mkt.databases.app (importable, shipped with mkt-databases) holds the
reusable, framework-agnostic visualization logic:
sequences.py — SequenceAlignment (kinase info + aligned sequences,
dict_align).schema.py — StructureConfig (abstract) + the StandardConfig /
StandardConfigChoice config enums; uses dependency injection (takes a
SequenceAlignment).structures.py — StructureVisualizer (CIF → Bio.PDB → PDB text,
get_highlight_data()).properties.py — PropertyTables.utils.py — create_structure_visualizer, validate_uniprot_indices.
missense_kinase_toolkit/app/ is the Streamlit front-end (not
pip-installable, deployed at https://mkt-app.streamlit.app). app.py /
visualizers.py import from mkt.databases.app and render with Streamlit.
Rule: logic in the package, UI in Streamlit
Put reusable data/structure/sequence logic in mkt.databases.app so both the
Streamlit app and the generate_pymol_files CLI can share it (the CLI imports
the same SequenceAlignment / StructureConfig / create_structure_visualizer
— see the pymol_viz skill). Keep Streamlit-specific widgets/layout in
missense_kinase_toolkit/app/. When adding a feature used by both, add it to
the package and call it from the front-end, not the reverse.
Dependency chain
mkt-schema → mkt-databases (incl. mkt.databases.app) → Streamlit app
The Streamlit app depends on both mkt-schema and mkt-databases; never make
mkt.databases.app import from the Streamlit app/ directory.
