OpenTFRaw

Part of the OpenMassSpec stack for mass spectrometry raw-file access.

Rust and Python reader for Thermo Fisher .raw mass spectrometry files. It is tested on format versions 63, 64 and 66, from LTQ ion traps through Orbitrap Astral and TSQ Altis (see Known issues and gaps and CORPUS.md).

Documentation: sigilweaver.app/opentfraw/docs

Known issues and gaps

Measured on the public test corpus (291 files from 56 PRIDE datasets).

• Scan metadata is wrong on 38 of 291 corpus files (13%). On these files the scan event is misread: MS level, polarity, analyzer, activation and precursor fields can be wrong, for example MS2 scans reported as negative-polarity MS1 in a positive-mode run. Peak m/z and intensity values are not affected. Affected: 30 of 52 Orbitrap Fusion Lumos files (PXD007683, PXD025230, PXD037285, PXD064117, PXD067311, PXD070648), 4 of 12 Q Exactive HF-X files (PXD021080), 2 of 2 Orbitrap Ascend files (PXD061791), 1 of 18 Orbitrap Eclipse files (PXD054530) and 1 LTQ file (PXD039587).
• Activation methods: CID, HCD, ETD and EThcD are identified. ETD and EThcD are read only from format version 66 files (no older ETD file is in the corpus). Other methods (UVPD, ECD, PQD and so on) decode as unknown, and mzML then has no dissociation-method term.
• Analyzer is not reported for Orbitrap Astral MS2 scans and for TSQ triple-quadrupole scans.
• SRM data: polarity is always reported as positive and isolation width as 0.7. Decoding of TSQ Vantage (format version 63) SRM files is unreliable: in PXD020246, some scans contain NaN values or peaks outside the scan's m/z range. 4 of its 35,098 scans have a flat-peak record that matches no known layout; they raise an error, and to_mzml(..., skip_unreadable=True) (Python) or the to_mzml example's --skip-unreadable flag writes the other scans and reports how many were left out.
• Profile spectra with an unrecognised calibration layout have NaN m/z values; mzML reports the count in opentfraw.unconverted_profile_bins.
• m/z accuracy follows the instrument's stored calibration. In FTMS MS1 scans of 223 corpus files, background ions with known masses (polysiloxanes, diisooctyl phthalate) are within 0.55 ppm of their theoretical m/z at the median and 4.1 ppm at the 95th percentile.
• Tested files: format versions 63, 64 and 66 only; positive polarity only in the corpus (two negative-mode Q Exactive Plus files back opt-in tests, see CORPUS.md); no GC-Orbitrap, LCQ-family, Orbitrap ID-X, Exactive, TSQ Quantis/Endura/Fortis or single-quadrupole files; no data from non-MS controllers (UV, PDA). Other versions and instruments may open but are untested.

Install

Prefer openmassspec-io with the thermo feature/extra unless you need this parser standalone (minimal dependencies, or building your own abstraction) - the umbrella gives you format auto-detection, mzML conversion, and Arrow streaming across all wired-in vendors for free:

cargo add openmassspec-io --features thermo

pip install openmassspec[thermo]

Standalone:

Rust:

cargo add opentfraw

Python:

pip install opentfraw

Quickstart

Rust:

use std::{fs::File, io::BufReader};
use opentfraw::RawFileReader;

let raw = RawFileReader::open_path("sample.raw")?;
// Buffer the scan reads: an unbuffered File is about 11x slower.
let mut file = BufReader::new(File::open("sample.raw")?);
let info = &raw.run_header.sample_info;
for scan_num in info.first_scan_number..=info.last_scan_number {
    let peaks = raw.read_scan_peaks(&mut file, scan_num)?;
    println!("scan {scan_num}: {} peaks", peaks.len());
}

Python:

import opentfraw

raw = opentfraw.RawFile("run.raw")
mz, intensity = raw.peaks(3)
print(raw.scan(3)["filter_string"])
raw.to_mzml("run.mzML")

See the docs site for the full guide, format specification, and API reference.

Used by

License

Apache-2.0. See LICENSE.

The format specification was developed by binary analysis of public mass-spectrometry datasets (PRIDE accessions). See CORPUS.md and ATTRIBUTION.md.