Bruker flex MALDI-TOF (fid + acqus)
flexControl, the acquisition software of Bruker’s MALDI-TOF instruments (autoflex, ultraflex, microflex, MALDI Biotyper, rapifleX), writes each spectrum as a directory holding fid (the summed detector signal) and acqus (its parameters). OpenReadout reads every spectrum below the path it is given, with m/z from the time-of-flight calibration in acqus. Bruker publishes no description of these files. The reader follows OpenMS’s XMASS reader (BSD-3-Clause), read as documentation, and was checked against pyOpenMS run on the same files. Provenance: docs/provenance/bruker-flex.md. Reader: crate openreadout-bruker-tims, module flex, format id bruker-flex.
Layout
A run is a directory tree <run>/<spot>/<acquisition>/<kind>/, for example 0_A4/1/1SLin/:
<spot>names the target position (0_A4);<acquisition>numbers the acquisitions of that spot (1,2);<kind>says how the spectrum was acquired:1SLinlinear,1SRefor1Refreflector, and<precursor m/z>.LIFT/1SRefa LIFT fragment spectrum.
A spot directory (is_spot) holds fid and an acqus (or acqu) whose first lines name XMASS (##TITLE= XMASS Parameter file or ##ORIGIN= XMASS, Bruker-Daltonics). flexAnalysis adds pdata/1/ with the processed spectrum 1r and its procs. The reader does not read pdata.
Detection. The input is a spot directory, its fid or acqus, or any directory above spot directories. find_spots walks at most 7 levels down and skips pdata; detection looks at no more than 400 directories. A directory above spots counts only when each of its subdirectories holds a spot (hidden folders, __MACOSX and pdata aside), so a folder that also holds other data is not taken for a flex run. Bruker NMR (TopSpin) directories also hold fid and acqus, but their acqus does not name XMASS, so the NMR reader leaves flex directories alone and the flex reader leaves NMR directories alone.
Parameters
acqus is JCAMP-DX style text: ##$KEY= value lines (some keys use ##. or no prefix), strings in angle brackets, arrays as (0..n) followed by value lines. The reader keeps every key, with its first line of text for arrays. Keys it uses:
| key | meaning |
|---|---|
TD |
number of points in fid |
BYTORDA |
byte order of fid: 0 little-endian, 1 big-endian |
DW, DELAY |
time per point and time of the first point (ns) |
ML1, ML2, ML3 |
calibration constants (below) |
IONIZATION MODE |
LD+ positive ions, LD- negative ions |
AQ_mod, SPType |
acquisition mode: in the corpus AQ_mod 1 is linear, 2 reflector, and SPType 2 marks a LIFT spectrum |
Parent |
the LIFT precursor m/z |
PATH |
the spot directory as the instrument wrote it |
NoSHOTS, INSTRUM, InstTyp, FCVer, AQ_DATE, CMT1, SPOTNO, PATCHNO |
laser shots, the instrument PC’s host name, a numeric instrument type, the flexControl version, the acquisition time, a comment, the spot name |
acqus does not name the instrument model.
Points and m/z
fid is TD signed 32-bit integers in the BYTORDA byte order. Point i has the flight time tof = DW · i + DELAY (ns). With b = √(10¹² / ML1) and c = ML2 − tof, s = √(m/z) solves ML3 · s² + b · s + c = 0:
ML3 = 0:s = −c / b;- otherwise
s = (√(b² − 4 · ML3 · c) − b) / (2 · ML3);
and m/z is s² (Calibration, mz). A spot whose acqus has no usable DW, DELAY, ML1 or ML2, or a non-positive ML1, is listed with a note and its points cannot be read.
LIFT spectra. A LIFT fragment spectrum gets m/z from the same constants. flexControl stores the fragment calibration in NTBCal, whose layout is not public, and the reader does not apply it, so fragment m/z can differ from flexAnalysis’s. notes and each such spectrum’s extra.mz_note say so.
Spectra
The run is one list of spectra, one per spot directory in path order. Each spectrum is a profile (centroided false), MS level 1, or 2 for LIFT with the precursor from Parent; polarity from IONIZATION MODE. native_id is the spot directory relative to the input. Intensities are 32-bit floats, exact up to 2²⁴.
| key | meaning |
|---|---|
spectra[0].extra.modes |
spectra per acquisition mode (linear, reflector, LIFT, unknown) |
spectra[0].extra.total_points, instruments, instrument_types, software_versions |
points in all spectra, INSTRUM, InstTyp and FCVer values |
spectra[0].extra.spots |
the first 1000 spots: path, spot, mode, points |
spectrum extra.spot, mode, points |
spot name (SPOTNO, else PATCHNO, else the path), acquisition mode, point count |
spectrum extra.shots, instrument, instrument_type, software_version, acquired, comment |
NoSHOTS, INSTRUM, InstTyp, FCVer, AQ_DATE, CMT1 |
spectrum extra.calibration |
DW, DELAY, ML1, ML2, ML3 |
spectrum extra.processed |
pdata/1/1r when flexAnalysis wrote a processed spectrum |
spectrum extra.mz_note |
LIFT spectra: the fragment calibration is not applied |
The acquisition mode comes from the directory names, else from the names at the end of PATH, else from AQ_mod.
vendor (info --view full) lists the first 200 spots with every acqus key (spots: path, acqus; spot_count).
check
check --headers-only parses every acqus (no_calibration, too_many_spots past 200,000 spots). check also reads every fid and computes its m/z (short_fid when fid holds fewer points than TD, bad_spectrum).
Vocabulary
| identifier | meaning |
|---|---|
BrukerFlexReader |
the flex FormatReader |
FLEX_FORMAT_ID |
bruker-flex |
FlexDataset |
an open run; spectrum returns one spectrum with its points |
Params, parse, get, is_xmass, Calibration, from, mz, is_spot, find_spots, dw, delay, ml1, ml2, ml3, map |
parameter file and calibration (sections above) |
modes, total_points, instruments, instrument_types, software_versions, spots, path, spot, mode, points, shots, instrument, instrument_type, software_version, acquired, comment, calibration, processed, mz_note, acqus, spot_count |
JSON keys (tables above) |
no_calibration, too_many_spots, short_fid, bad_spectrum |
check finding codes |
linear, reflector, LIFT, unknown |
acquisition modes |
How this reader was derived, file by file: provenance log.