Reproducible engine comparisons for andes against the open-source field. Each benchmark documents the datasets, the exact per-engine parameters, and the FDR methodology so the numbers can be regenerated.
Canonical public benchmark — andes vs Java MS-GF+ vs Comet:
| Benchmark | Date | Engines | Datasets |
|---|---|---|---|
| Public benchmark | 2026-06-15 | andes (top-1 + --chimeric), Java MS-GF+, Comet |
Astral (HCD high-res), TMT a05058 (CID low-res), UPS1/PXD001819 (CID low-res) |
Every engine is re-scored through one uniform Percolator (3.7.1, --seed 42 -Y),
and FDR honesty is verified independently with a 1:1 entrapment search (true FDP
≈ 1% at the nominal 1% q-value). Per-engine configuration files live under
configs/; reproducibility scripts under scripts/.
Internal / superseded development reports
Early multi-engine validation runs (include MSFragger / Sage / ProSE for development context), kept for reproducibility but superseded by the public benchmark above: