SynEPD

SynEPD is a hierarchical database of polar organic reaction mechanisms. It combines curated atom-mapped reactions, the POLAR taxonomy, reaction-center (RC) and mechanistic-center (MC) templates, ITS graphs, and ordered electron-pushing diagram (EPD) arrows in SQLite and a web explorer.

• Web explorer
• Zenodo releases
• Documentation

Dataset

Item	Count
Reactions	1,926
Molecules	2,277
RC templates	1,521
MC templates	1,540
EPD arrows	8,129
Taxonomy classes	939
Accepted RXNO/MOP links	218

An RC describes changed bonds. An MC extends its RC with the transition edges required by the curated electron flow. In the current release, 284 MC templates (18.44%) differ from their RC and 229 (14.87%) structurally extend it.

The corpus covers closed-shell, paired-electron mechanisms. Concerted reactions are included when their electron movement is expressible with two-electron arrows. Radical and single-electron-transfer mechanisms are excluded because the current vocabulary has no fishhook-arrow representation.

Quick Start

Create the Conda environment and install the package:

conda env create -f env.yaml
conda activate synepd
python -m pip install -e ".[dev]"

Start the local explorer:

./run_server.sh

Then open http://127.0.0.1:8000/. Versioned API routes are available under /api/v1; /api remains as a compatibility alias.

Query from Python

from synepd.query import SynEPDQuery

with SynEPDQuery("data/epdb.sqlite") as query:
    result = query.epd("CC[O-].[NH4+]>>CCO")
    for arrow in result.get("arrows", []):
        print(
            arrow["arrow_index"],
            arrow["arrow_type_code"],
            arrow["source_atoms"],
            "->",
            arrow["target_atoms"],
        )

The same interface supports molecule, taxonomy, arrow-sequence, RC-template, ITS/RC/MC, and RXNO/MOP queries. SynEPDQuery.from_release(...) resolves a tagged release through the zenodo, github, or auto sources.

Build and Validate the Data

Build the release database:

PYTHONPATH=. python -m synepd.construct.build_release_db

Rebuild or check the external RXNO/MOP linkage:

python scripts/build_rxno_mapping.py --obo data/rxno.obo
python scripts/build_rxno_mapping.py --check --obo data/rxno.obo

Verify the checked-in release artifact:

python -m synepd.construct.release_manifest data/epdb.sqlite \
  --verify data/release-manifest.json

Core release files are:

Path	Purpose
data/polar.json	Curated production reaction records
data/hierarchy.md	POLAR taxonomy source
data/epdb.sqlite	Built read-only release database
data/rxno_crosswalk.tsv	SynEPD-to-RXNO/MOP linkage
data/release-manifest.json	Version, counts, and checksums
synepd/web/static/data_arch.tex	Editable database diagram source

SQLite contains the reaction, molecular, taxonomy, RC, MC, ITS, and EPD core. RXNO/MOP mappings remain versioned TSV/RDF artifacts and are resolved on demand instead of being duplicated in SQLite. Narrative curation notes and internal named-reaction relations are not exposed through the public schema.

Development Checks

python -m black --check --workers 1 synepd test
./lint.sh
./pytest.sh
python -m synepd.precheck.epd_verification --strict

The EPD command uses strict ordered electron-resource replay by default. It checks every intermediate state, charge propagation, locality, and the mapped product endpoint. Pass --legacy-editor only for compatibility comparisons with the historical endpoint-only check.

For a multi-worker deployment, run ./run_server.sh run --no-reload --workers 4 behind a reverse proxy. The bundled SQLite release is opened read-only; PostgreSQL is supported for horizontally scaled or write-heavy deployments.

License and Acknowledgment

The software is licensed under Apache License 2.0. Curated release data is distributed under CC BY 4.0 where stated in its metadata. See LICENSE.

This project received funding from the European Union's Horizon Europe Doctoral Network programme under Marie Skłodowska-Curie grant agreement No. 101072930 (TACsy).