MolProp Toolkit stays stable as the “table compiler” and schema authority. MolScope Server is the fast-moving companion that adds a Streamlit GUI, interactive Plotly dashboards, and higher-level medchem workflows while keeping existing CLI pipelines intact.
The platform supports two entry paths that converge on the same artifact: a MolProp results table and a run folder with logs, plots, and optional HTML reports/picklists. You can start from SMILES (calculator mode) or start from an existing results table (analysis-only mode).
Upload .smi → run molprop-calc-v5 → optional report/picklists → download a zip bundle.
Upload .parquet/.csv → PCA/UMAP dashboards → optional report/picklists → download a zip bundle.
git clone https://github.com/kelokely/molprop-platform.git
cd molprop-platform
pip install -e ".[dev,web,viz]"
Installs the Streamlit UI and visualization stack.
# installs MolProp Toolkit from GitHub (until core is on PyPI)
pip install -e ".[core]"
Enables calculator mode in the web app by providing molprop-calc-v5.
Start the Streamlit app and choose a workflow tab inside the UI.
molscope-server
If you installed the core toolkit, the app can run calculators from uploaded SMILES; otherwise it runs in analysis-only mode.
Generate a PCA/UMAP dashboard from an existing results table.
molscope-visualize results.parquet -o viz --method umap
Outputs an interactive HTML scatter plot plus a projection CSV.
MolProp Toolkit and MolScope Server are connected by a stable interface: the results table schema. Toolkit produces tables; platform consumes them and creates additive artifacts. This keeps existing pipelines intact and makes Milestone B tooling easy to build as a table consumer.
# 1) make a MolProp table
molprop-calc-v5 library.smi -o results.parquet
# 2) visualize and share
molscope-visualize results.parquet -o viz --method pca
Quickstart runs a complete happy path and explains the install options.
Tools lists CLI commands shipped in the platform.
Toolkit docs is the schema and column authority.
Table-first, additive outputs, explicit provenance, optional dependencies behind extras.