File Formats

Input: MGF

msfiddle accepts tandem mass spectra in .mgf format. Four fields are required per spectrum; all others are ignored.

Field	Description
`TITLE`	Unique spectrum identifier, propagated to the `ID` column in output.
`PRECURSOR_MZ`	Observed precursor m/z.
`PRECURSOR_TYPE`	Adduct type (e.g. `[M+H]+`, `[M-H]-`). See supported types below.
`COLLISION_ENERGY`	Collision energy in eV.

Example:

BEGIN IONS
TITLE=EMBL_MCF_2_0_HRMS_Library000529
PEPMASS=111.02016
CHARGE=1-
PRECURSOR_TYPE=[M-H]-
PRECURSOR_MZ=111.02016
COLLISION_ENERGY=50.0
41.0148 0.329893
68.0258 0.402906
111.0203 100.0
END IONS

Supported precursor types:

[M+H]+, [M+2H]2+, [M+Na]+, [M-H]-, [M+H-H2O]+, [M-H2O+H]+, [2M+H]+, [2M-H]-, [M+H-2H2O]+, [M+H-NH3]+, [M+H+NH3]+, [M+NH4]+, [M+H-CH2O2]+, [M+H-CH4O2]+, [M-H-CO2]-, [M-CHO2]-, [M-H-H2O]-

Output: msfiddle CSV

One row is produced per input spectrum.

Column	Description
`ID`	Spectrum identifier from the MGF `TITLE` field.
`Y Pred`	Raw TCN output: semicolon-separated atom-count vector (C, H, O, N, F, S, Cl, P, B, Br, I, Na, K).
`Mass`	Neutral monoisotopic mass derived from `PRECURSOR_MZ` and `PRECURSOR_TYPE`.
`Pred Formula`	Top formula from the TCN model prior to refinement.
`Pred Mass`	Monoisotopic mass of `Pred Formula`.
`Pred Atom Num`	Total atom count predicted by the model.
`Pred H/C Num`	H/C ratio predicted by the model.
`Running Time`	Wall time per spectrum in seconds (prediction + refinement).
`Refined Formula (k)`	The k-th best refined formula (0-indexed), ranked by rescore score. `None` if fewer than k+1 candidates were found.
`Refined Mass (k)`	Monoisotopic mass of `Refined Formula (k)`.
`Rescore (k)`	Rescore model confidence score for `Refined Formula (k)` (0–1; higher is more confident).

The number of ranked columns is set by top_k in the configuration file (default: 5).

Input: BUDDY CSV (optional)

Required columns when using --buddy_path:

Column	Description
`ID`	Spectrum identifier matching the MGF `TITLE` field.
`Adduct`	Precursor type string.
`Pred Formula (1–5)`	Top-5 candidate formulas from BUDDY.
`BUDDY Score (1–5)`	Confidence scores; candidates below the configured threshold are excluded.

Input: SIRIUS CSV (optional)

Required columns when using --sirius_path:

Column	Description
`ID`	Spectrum identifier matching the MGF `TITLE` field.
`Pred Formula (1–5)`	Top-5 candidate formulas from SIRIUS.
`Pred Adduct (1–5)`	Predicted adduct for each candidate.
`SIRIUS Score (1–5)`	Log-likelihood scores; candidates below the configured threshold are excluded.