Viewer and editor for UV-Vis, FTIR and fluorescence spectra: load, overlay and compare multiple spectra together, with an interactive cursor on the plot and a set of processing tools — scattering correction (Rayleigh/Mie), adaptive baseline, scaled subtraction between spectra, averaging, smoothing (boxcar or Savitzky-Golay) and multi-peak deconvolution (Lorentzian/Gaussian/pseudo-Voigt).
| Extension | Type | Notes |
|---|---|---|
.dsp |
UV-Vis (binary) | native parser |
.sp |
FTIR PerkinElmer (binary) | native parser (no external tool required) |
.csv |
UV-Vis or FTIR | type auto-detected from the X-axis values |
.csv |
UV-Vis (BMG plate reader) | absorbance spectra per well; detected from the header — see BMG plate-reader CSV |
Support for other instrument/vendor-specific formats can be added on request — open an issue with a sample file.
A .csv exported by a BMG microplate reader (the sample files came from a BMG Omega, with
the data folder ...\BMG\Omega\User\Data in the header) holding absorbance spectra, one per
well. It is recognised by Test run no.: in the first lines and read by _leggi_csv_bmg().
Tests: tests/test_bmg_reader.py.
Text file, ; separator, CRLF line ends, . as decimal separator in the files seen so far. One
row of metadata per line, then the table:
User: USER;Path: C:\...\BMG\Omega\User\Data;Test run no.: 459
Test name: 350_1000_sp;Date: 03/07/2025;Time: 17:02:38
<- blank line
ID1: <experiment name>; <- line 4
Absorbance spectrum
<- blank line
Raw Data (Abs Spectrum)# <- or "Blank corrected based on Raw Data (Abs Spectrum)#"
<- blank line
Well; Content; 1 - 1; 1 - 1; ... <- one "1 - 1" per wavelength (ignored)
;Wavelength [nm];350;351;352;...;1000; <- 651 points, 1 nm step in the sample
A01;Blank B;0.322;0.320;... <- one row per well: id; content; one value per wavelength
A02;Sample X1;0.384;0.381;...
Every table row — the Well; Content row, the Wavelength row and the data rows — ends with a
trailing ;, so each has 2 + number of wavelengths + 1 fields (654 for 651 wavelengths).
The metadata lines have no fixed position apart from being above the ;Wavelength [nm] row:
the reader looks for that row instead of counting lines.
The same run can be exported twice:
- Raw data — every well has its values, including the
Blankwells. - Blank corrected — the
Blankrows are present but empty (all fields blank), and each sample issample − mean of all the blank wells(the mean over the blank wells of the whole plate, not the blank in the same row: for sampleA02the blankA01alone gives a different result). Checked on one export (3 samples, 3 blanks, 350–1000 nm): subtracting the mean of the raw blanks reproduces the software's corrected values to within 0.001, i.e. the 3-decimal rounding of the file. So the raw file is enough to obtain the corrected spectra.
- Each well with data becomes one spectrum, named
<file name>_<well>(e.g.run_A02); the well content (Sample X1), experiment name (ID1), test name and run number, date and time, the "Raw Data / Blank corrected" line and the instrument appear in the metadata panel. - Spectra are typed UV-Vis (X axis in nm).
- Raw file with
Blankwells: a dialog asks whether to subtract the mean of the blanks. Yes loads only the samples, already corrected; No loads samples and blanks as they are. - Blank-corrected file: the empty blank rows are skipped, nothing is asked.
- Empty cells are dropped; a decimal comma is accepted; the file is read as
latin-1like the other CSV readers. - A well whose content starts with
Blankis treated as a blank, anything else as a sample. If a file contains only blanks they are loaded as ordinary spectra.
Only one export has been inspected, with a single plate, a single measurement per well and only the
contents Sample and Blank. Not covered, because they were never seen: other well types
(standards, controls — they would be loaded as samples), several plates or repeated cycles in one
file, fluorescence or luminescence exports (the header says Absorbance spectrum), and a Time
or Date written in another locale format (shown as found). If a file does not load, send the first
10–15 lines of the export — the header and two data rows are enough to adapt the reader.
numpy, pandas, matplotlib, tkinter (standard library); tkinterdnd2 is optional and
enables drag-and-drop file loading; scipy is optional and enables the constrained refinement
step in the scattering correction auto-fit (a plain least-squares fallback is used without it).
PlotStyleKit is an optional sibling repo (cloned as
../PlotStyleKit next to this project) providing the shared Origin-like plot style and the
standalone figure editor behind File → Save Figure (pickle) / File → Edit Figure...; without
it those two menu items are unavailable and the app uses default matplotlib styling.
pythonw LabSpectrumManager.pyw [spectrum_file]
Three panels side by side: a data table of the merged spectra on the left, a Matplotlib graph with an interactive crosshair cursor in the center, and a list of loaded spectra (multi-select) with metadata on the right — which scrolls vertically, since some tool panels (Deconvolution) can need more room than the window has. The X/Y readout that follows the cursor picks its corner automatically to avoid the plotted curves, and can be dragged anywhere on the graph.
The data table's columns (one per spectrum, plus X) can be resized by dragging their border, and the table scrolls both ways once there are more columns or rows than fit. Select rows (click, Shift/Ctrl to extend, Ctrl+A for all) and Ctrl+C — or right-click — to copy them, header included, tab-separated, ready to paste into Excel/Origin.
Right-click a spectrum in the list to rename it or to move it (up, down, top, bottom; several
selected spectra move together). The list order is the order of the graph colours and of the columns in
File → Export CSV. Rename and move are not available while an operation panel is open.
Any spectrum can be hidden from the graph (Hide Selected / Show Selected, or right-click in the
list) without removing it — it stays in the list (greyed out), data table and metadata, just off
the plot. Closing the app (window close button or File → Exit) asks for confirmation if there
are calculated results — averages, fits, corrections, trims, pasted data — that haven't been
exported yet with File → Export CSV or saved with File → Save Session....
File → Save Session... saves everything currently loaded — every spectrum, including ones
derived from a processing tool (scattering correction, baseline, smoothing, fit, trim, average,
...) — to a single .lsmsession file, so the work can be resumed later instead of redone step by
step. File → Open Session... reloads it, replacing what's currently loaded. Each processing
panel must be closed (Apply or Cancel) before saving, with one exception: Deconvolution can be
saved mid-fit — peaks, fit range, profile and offset not yet applied — and reopening the session
restores the panel to that exact state so the fit can be resumed.
File → Save Figure (pickle) saves the current graph as a live, re-editable Figure object
(not a raster image); File → Edit Figure... opens it directly in PlotStyleKit's figure editor
for titles, axis labels, per-line styling, legend placement and publication-size PNG/SVG/PDF
export — see Dependencies.
- Scattering correction — estimates and subtracts a baseline of the form
VS · (λ/1000)^-P − M·λ + Offset, with a constrained auto-fit (SLSQP) over the whole spectrum by default: the power term models Rayleigh (P≈4, small particles) to Mie (P down to 1, large particles) scattering, the linear term covers residual baseline drift unrelated to scattering and is fixed at zero unless unlocked. Theory and operating procedure inscattering_correction_theory.md. - Linear baseline (FTIR) — a two-anchor baseline, draggable on the graph, for FTIR spectra.
- Adaptive baseline — a lower envelope via iterated moving averages with a minimum constraint, adjustable window via a slider (from "flat" to "hugging the signal").
- Scaled subtraction — subtracts a reference spectrum B from a spectrum A with an adjustable
coefficient
k(Result = A − k × B), useful for removing a solvent/blank contribution. - Average of several selected spectra.
- Smoothing — repeated moving average (boxcar, ~Gaussian) or Savitzky-Golay filter (local polynomial fit, better preserves peak height and shape).
- Derivative — 1st or 2nd derivative via Savitzky-Golay (exact derivative of a local polynomial fit, in real X units), with the same adjustable smoothing/polynomial-order controls as Smoothing, useful for resolving overlapping bands or removing a sloping baseline.
- Shift / Scale — vertical translation and/or multiplication by an arbitrary factor
(
Y_new = Y × Factor + Offset) of one or more selected spectra, with live preview; the result is stored as new_ssspectra and the originals are left untouched. - Deconvolution — interactive multi-peak fit (click to add/remove a peak) with a Lorentzian, Gaussian or pseudo-Voigt profile; the fit is restricted to a window settable by dragging its two boundary lines or typing exact values, same as Trim below.
- Trim — crops a spectrum to a window (drag the two vertical lines, or type exact start/end values) and stores the selected portion as a new, independent spectrum, leaving the source trace untouched; the window edges always snap to a real data point.
LabSpectrumManager.pyw— main application (a singleLabSpectrumManagerclass)old version and side projects/— earlier versions (v0-v2) and side projects (scattering.pyw,spc_plotter.pyw) kept as historical referencetest_sp_reader.py— test for the native.spparser (FTIR PerkinElmer)tests/test_bmg_reader.py— unit tests for the BMG plate-reader CSV reader (synthetic files; run withpython -m unittest discover -s tests -v)