synrecon 0.0.1

Utility for reaction modeling using graph grammar

SynRecon

Reconstruct molecular formulas from exact (monoisotopic) mass.

Given a neutral exact (monoisotopic) mass, SynRecon searches for a non-negative integer composition over a chosen element set that matches the mass within a configurable tolerance (default 1e-5 Da). Enumeration is performed using integer scaling with a fast DFS enumeration and an optional ILP backend. Scope: basic single best formula; optional chemistry filters (DBE ≥ 0, even-electron rule, nitrogen rule) can be enabled to tighten results.

Highlights

• Integer-weight DFS enumeration for raw candidate formulas.
• Optional chemical filters: DBE, even-electron rule, nitrogen rule.
• Optional simple isotopologue expansion (e.g. D for H).
• Small, dependency-light core; optional ILP backend for constrained searches.

---

Install

pip install synrecon

Quick examples

Python API (recommended)

from synrecon.solver import solve_formula, enumerate_formulas

# single best solution (or None)
sol = solve_formula(
    180.063388104,
    elements=["C", "H", "O"],
    abs_tol_da=1e-5,
    enforce_dbe=True,
)

if sol is None:
    print("No solution found within tolerance.")
else:
    print("Best formula:", sol.formula)
    print("Calculated mass:", sol.mass_calc)
    print("Absolute error (Da):", sol.abs_error)
    print("DBE:", sol.dbe)

# enumerate several candidates (more verbose)
candidates = enumerate_formulas(
    180.063388104,
    elements=["C", "H", "O"],
    abs_tol_da=1e-5,
    enforce_dbe=True,
    max_candidates=20,
    sort_by="abs_error",
)

print("\nTop candidates:")
for s in candidates[:10]:
    print(f"{s.formula:12s} mass={s.mass_calc:.6f} err={s.abs_error:.6e} dbe={s.dbe}")

CLI

synrecon solve --mass 180.063388104 \
              --elements "C H O" \
              --tol 1e-5 \
              --dbe \
              --allow-isotopes \
              --labels "D T"

Contributing

• Tieu-Long Phan

Publication

SynRecon:

License

This project is licensed under MIT License - see the License file for details.

Acknowledgments

This project has received funding from the European Unions Horizon Europe Doctoral Network programme under the Marie-Skłodowska-Curie grant agreement No 101072930 (TACsy -- Training Alliance for Computational)