Free LCMS Data Analysis Software That Actually Works
You process the run, you get the mzML file, and now you need to find peaks, align them across samples, and figure out which features actually matter. Most people go straight to whatever their lab mates use. The free options are decent if you know what you are doing, and they are not if you don't. Here is a breakdown of the tools I actually reach for, what each one is good at, and where they quietly fail. XCMS remains the workhorse for anyone doing untargeted LC-MS metabolomics. It runs inside R, which means you can script the entire preprocessing pipeline and repeat it exactly. The core functions — findPeaks, group, retention correction, and fillPeaks — handle the heavy lifting for peak detection, feature grouping, and retention time alignment. The default parameters will almost never be optimal for your data, so you need to tune the peakwidth, snthresh, and ppm values based on your instrument's actual resolution. I once spent three days fighting XCMS because my early eluting peaks were being split into multiple features. The fix was lowering the snthresh and switching to the centWave method instead of the default matchedFilter. It cut false feature splitting by about forty percent on my samples. MS-DIAL is the other tool I come back to, especially for lipidomics and when you need annotation against public spectral libraries built in. It does chromatogram extraction, peak detection, and isotope pattern filtering all in one GUI. The library search function supports NIST, HMDB, and you can import your own curated spectra. Where it gets complicated is the deisotoping and adduct grouping — MS-DIAL tries to do everything automatically, and that automation sometimes merges features that should stay separate. My workaround was to run the initial processing with default settings, export the feature table, then manually review any feature groups that had more than five isotopic clusters. It takes time but it prevents downstream statistics from being skewed by mis-grouped features.
MZmine 2 gives you a modular pipeline where every step is transparent. You can select different algorithms for peak detection (deconvolution, ADAP, bin-based), choose your own grouping method, and interpolate missing values however you want. The advantage is control. The disadvantage is that you have to understand what each module does before it will give you trustworthy results. I prefer MZmine when I am working with complex matrices like plasma or soil extract where baseline drift and co-eluting interference are constant problems. The binning-based peak detection handles noisy baselines better than most, but it requires you to set the bin width correctly or you lose low abundance features entirely. A bin width of two to three points across your chromatographic peak is usually safe for standard C18 runs. OpenMS is less of a point-and-click tool and more of a framework. If you are comfortable with Python or command line workflows, it offers incredibly fine-grained control. FeatureFinderMetabo, MapAlignerIdentification, and QFeatureConverter are among the modules I use most. The learning curve is steep. Setting up a complete processing chain takes longer than opening MS-DIAL, but once it is scripted, you can run fifty samples through it without touching anything. The real value shows up when you need reproducibility for regulatory submissions or when your lab generates data at scale. One thing OpenMS does better than most: it handles empty scan removal and charge state filtering without you having to write custom code. ProteoWizard is not a full analysis suite. It is primarily a converter. You feed it raw files from Thermo, Agilent, Waters, or Sciex, and it spits out mzML or mzXML. Almost every other tool on this list depends on ProteoWizard to read vendor formats. Keep it installed. The msConvert command line tool lets you batch convert entire folders, which matters when you have three hundred injections to process.
Skyline deserves a mention even though it is targeted, not untargeted. If your goal is quantification of known compounds — pharmaceutical impurities, biomarker validation, pharmacokinetics — Skyline is faster and more accurate than any untargeted pipeline. You set up transitions, define your standards, and it handles integration, calibration curves, and quality control flags automatically. The free version is fully functional. The only limitation is that you cannot discover new features with it. It is a targeting tool, and it is excellent at that. There are tradeoffs with every free option. XCMS struggles with large datasets above two hundred samples unless you have a lot of RAM. MS-DIAL's automatic annotation occasionally pulls in wrong matches from incomplete spectral libraries. MZmine's documentation is sparse and the UI feels dated. OpenMS requires programming knowledge. None of these tools will fix bad chromatography. If your peaks are merging or your retention time drift is extreme, preprocessing software will only mask the problem, not solve it. In those cases, revisit your method development before spending hours tuning software parameters. For most labs starting out, I recommend running the same batch through both XCMS and MS-DIAL and comparing the overlap. Features that appear in both are more likely to be real. Features unique to one pipeline deserve manual inspection. That comparison step usually takes about twenty minutes per batch and catches errors that would otherwise show up weeks later during statistical analysis.
Get the Full Details
