Skip to content

Troubleshooting

Use this page to diagnose common installation, input, extraction, and export problems. If the problem is not listed, open an issue with the steps required to reproduce it.

Install / launch

ProblemCauseFix
command not found: leafInstall location not on PATHuv tool update-shell (uv) or add ~/.local/bin to PATH (pip)
Unsure whether LEAF installed correctlyMissing package, native extension, reader backend, or Web UI bundleRun leaf doctor; add --strict when optional checks should fail setup scripts
Port 18008 already in useAnother process is on the portleaf webui run --port 18009
pythonnet errors on WindowsMissing .NET 8.0Install .NET 8.0 SDK
Browser shows "Cannot connect"LEAF crashed or terminal closedRe-run leaf webui run; check the terminal for errors
LEAF starts but no logo / stylesBrowser cached an old buildHard-refresh with ⌘⇧R (Mac) or Ctrl+Shift+R (Win/Linux)

Input files

ProblemCauseFix
RAW file fails to loadUnsupported instrument firmwareTry opening the file in Thermo Xcalibur first; if it works there, report it
"No samples found in folder"Folder has no supported input files, or files mix formatsUse .raw, .mzml, or .mzml.gz and keep one format per run.
Sample names look unexpectedAuto-name extraction parsed the filename incorrectlyToggle "Organize names" off to use the raw filename
Blank files included anyway"Skip blanks" only matches the word "blank"Rename blank files to include "blank", or untoggle "Skip blanks" and remove them after extraction
RAW file fails to load on macOS / LinuxSEED reader hit an unsupported instrument firmwareSwitch to a Windows machine and try the dotnet backend (leaf targeted ./samples ./compounds.csv ./outputs --backend dotnet); if it still fails, report it
Unsure whether a CSV / folder is validInput preflight not run yetleaf validate ./compounds.csv ./raw-folder; add --strict to treat warnings as failures

Compound list

ProblemCauseFix
"Missing column: Metabolite"CSV header does not match a recognized compound-name fieldUse one of: Metabolite, Compound, Name
Compounds show as "invalid formula"Formula has a typo or unsupported elementCheck the formula syntax (e.g., C6H12O6, not C₆H₁₂O₆); only standard atoms allowed
All compounds say "no peaks detected"Wrong polaritySwitch polarity (NEG ↔ POS) and re-run
Some compounds are detected and others are notMass tolerance is too tight or the expected RT is inaccurateLoosen mass tolerance to 10 ppm; widen RT window to 1.0 min

Extraction

ProblemCauseFix
"backend is unavailable" errorThe selected reader backend is not installed or not detectedRun leaf doctor to check backend status. For SEED: install the seed wheel. For .NET: install .NET 8 runtime.
Backend disabled in the web UILEAF detected that the backend cannot run on this systemHover the disabled option for details; install the missing dependency or switch to an available backend
MS² extraction ignores backend choice.NET RawFileReader does not support the MS² extraction surfaceExpected behavior — MS² auto-routes to the SEED (Rust) backend
Extraction is unusually slowLarge dataset, slow storage, or non-default reader backendUse the Rust backend in Settings → Advanced and keep input files on fast local or network storage
Out-of-memory crashToo many samples in one batchProcess in smaller batches (50 files at a time)
Floating button stuck blueJob hung — usually a corrupt RAW fileCancel the job, remove the suspect file, re-run
Job fails silentlyDisk full or write permissions issueCheck disk space and the configured Storage path

Peaks and quality

ProblemCauseFix
Many "Poor" verdictsMass tolerance too loose, picks up noiseReduce mass tolerance to 5 ppm
Peaks at wrong RTRT drift between samplesIncrease RT Window to 1.0 min; or use RT alignment
EIC chart is emptySelected sample has no signal for the compoundCheck the compound's mass and adduct; try the m/z manually in Xcalibur
Automatic peak picking misses a visible peakPeak picker is conservative for the current chromatogramWiden the RT window, then manually drag-select on the EIC chart if needed
RT check shows huge outliersSample truly has different RT (different LC method?)Verify the sample is from the same method as others

Visualizations

ProblemCauseFix
Volcano plot shows errorMore or fewer than 2 sample groupsGroup samples into exactly 2 groups in the sidebar
Heatmap is all one colorForgot to normalizeSet Normalization to Z-score in the sidebar
PCA shows samples bunchedFew principal components explain little varianceIncrease the number of PCs displayed; check sample QC first
Network is emptyEdge threshold too strictLower the edge threshold slider

Export

ProblemCauseFix
.msd file won't reopenSaved with a much newer LEAF versionUpdate LEAF — see GitHub Releases
Need to check what's inside a result archiveFile came from another run or collaboratorleaf inspect ./result.msd
CSV missing isotopologues"Include isotopologues" was offRe-export with the option enabled
CSV has scientific notationExcel auto-converts large numbersOpen in a text editor or import as text in Excel

Hosted (MINT) mode

Hosted LEAF through MINT is under development and not enabled for general use yet. Use the wheel + CLI install path for current analyses.

Still stuck?

  1. Check the terminal (desktop) or browser dev console (F12 → Console) for error messages.
  2. Search GitHub issues for existing reports.
  3. Open a new issue with: LEAF version (leaf --version), OS, reproduction steps, and the error message.

LEAF is developed by the Morscher Lab.