
PFAS Analysis
Non-targeted annotation of known and novel PFAS structures in environmental and consumer samples. SIRIUS flags polyfluorination directly from your LC-MS/MS data, groups CF2 homologous series and proposes formulas and structures for PFAS that no target list or spectral library contains, without flooding the rest of your sample with false fluorine hits.
Why PFAS teams use SIRIUS
Which PFAS are actually in this sample, beyond the ones on your target list?
Routine PFAS methods measure a fixed set of targets, while the class spans thousands of structures. Replacement chemistries, production byproducts and transformation products appear in samples long before they appear on a monitoring list. Whatever your sample, the open question is the same, and a targeted panel cannot answer it.
SIRIUS answers it from the LC-MS/MS data you already acquire: it flags fluorinated features automatically, characterizes them without needing a library entry, and gives you the evidence to decide which candidates go to confirmation with standards.
Flag
Spot fluorinated features automatically during pre-processing, from mass defect, isotope pattern and homologous series, before any database search.
Trace
Follow precursors, byproducts and transformation products back to their source, across samples, sites and treatment steps.
Characterize
Propose formulas, compound classes and structures for PFAS that exist in no spectral library or structure database.
PFAS detection built into general small molecule annotation
SIRIUS is a general annotation tool: it proposes formulas, compound classes and structures for every feature in your sample, whatever its chemistry. PFAS detection is not a separate mode you switch on. It runs automatically as part of pre-processing, reads the evidence for polyfluorination directly from your data, and hands flagged features to the same annotation pipeline as everything else.
- No prior knowledge required: You don’t need a target list, a suspect list or a hand-tuned element set. SIRIUS recognizes polyfluorination from the mass defect and isotope pattern, and from CF2 homologous series in your LC-MS data.
- Fluorine only where the data supports it: High-fluorine formulas enter the de novo search only for flagged features. The rest of your dataset keeps a compact formula space, so computation stays fast and spurious fluorine annotations stay out.
- PFAS must win on the evidence: A flag widens the search, it does not decide the answer. PFAS and non-PFAS candidates are scored by the same fragmentation model, trained on fragments and losses from all compound classes.
A feature is only annotated as PFAS if that is objectively the best hypothesis,
while the rest of your dataset keeps its fast, standard annotation.
PFAS detection in full non-target datasets
Fluorine element prediction: a transformer-based network combines mass defect and isotope pattern to detect polyfluorinated compounds.
Homologous series: finds CF2 series, validates them by MS/MS similarity and passes PFAS predictions to all members, shown in a Kendrick mass defect view.
Dynamic search space: fluorine is enabled only for flagged features, so the rest of the dataset keeps fast, clean annotation.
PFAS filter and blank subtraction: show only PFAS-flagged features and hide background and instrument contamination.
Field example

Characterize novel PFAS that exist in no database
Molecular formula annotation across diverse compound classes, not just PFAS.
Custom suspect databases: import your own PFAS suspect lists and search them with SIRIUS, while falling back to PubChem when a better candidate exists there.
De novo structure generation proposes structures for PFAS absent from every database.
Structure Sketcher to modify a candidate by hand and re-score it instantly against the predicted fingerprint.
SIRIUS discoveries
Over 4,000 bioactives and hidden contaminants, including PFAS, annotated and semi-quantified in bee pollen supplements.
Environmental risk assessment & exposure biomonitoring
LC-MS/MS pre-processing with feature detection, alignment across many samples, adduct detection and quality metrics for prioritization.
Spectral library matching: in-house PFAS libraries and a curated public library from GNPS, MassBank and MsnLib, with identity and analog search.
One run, all contaminant classes: PFAS flagging runs alongside general small molecule annotation, so PFAS and co-occurring contaminants come from the same dataset.
High-risk prioritization: SIRIUS molecular fingerprints feed tools that predict toxicity and ionization efficiency, so high-risk features can be ranked before structure assignment.
Field example
594 human milk samples from Canada and South Africa were analyzed by LC-MS/MS, with SIRIUS supporting structure elucidation and candidates validated with standards. PFAS and phthalate metabolites appeared alongside parabens, including two parabens reported in human milk for the first time.
More SIRIUS discoveries
Extreme weather events can remobilize both current-use and legacy contaminants down to depths of 90 meters in fractured rock groundwater.
114 chemicals migrating from food packaging identified, including hazardous plasticizers and persistent non-intentionally added substances.
Tracking PFAS degradation & transformation products
Reaction Workflows: define custom chemical reactions, chain them into workflows and apply them to your custom structure databases to generate candidate transformation products.
Combined precursor and transformation product screening: search precursor and transformation product databases together with a broad structure database, so unrelated compounds are not forced onto a transformation product.
Substructure validation with analog hits: highlight the shared structural core between a candidate and related library spectra to check whether a proposed transformation explains the data.
Tutorial
A step-by-step walkthrough of non-targeted screening for transformation products of pharmaceuticals, pesticides and industrial chemicals. From generating a transformation product database to analysing LC-HRMS data in SIRIUS and validating the results.
A PFAS call you can defend
With regulators and the public watching, a false PFAS positive costs as much as a missed one. SIRIUS makes the evidence behind every annotation visible, so you can decide what goes to confirmation with standards.
Best hypothesis, not a PFAS bias
Flagged features are scored by the same model as every other compound learned across diverse classes.
Validated series, not mass coincidences
A repeating CF2 mass shift alone is not enough: series members must also share MS/MS spectral similarity.
COSMIC confidence score
Rank structure hits across a whole dataset; also for candidates one reaction away from the true structure.
Substructure validation
Link individual fragment peaks to parts of the candidate structure, and check shared substructures in analog library hits.
Library cross-checks
Identity and analog matches from in-house and public spectral libraries are shown next to the de novo results.
Built for monitoring campaigns with thousands of features
PFAS screening rarely means one sample. SIRIUS processes millions of queries every year and fits into the workflow you already have, from an interactive session on one run to fully automated pipelines across sites and seasons.
Vendor-neutral input
Import LC-MS runs in open formats and peak lists from other tools, with alignment across many runs.
GUI, CLI and API
Explore results interactively, batch-process from the command line, or run SIRIUS as a headless service in your own pipeline.
Computation presets
Start from predefined workflows for general annotation, database search only, or MS1-only data, and save your own.
High-risk prioritization
Use the SIRIUS molecular fingerprints to predict toxicity and ionization efficiency.
Integrations and export
Export summaries for downstream tools or connect SIRIUS with other software.
Which PFAS are in your samples?
Related solutions
Environmental and water analysis
Unknown pollutants and degradation products in water and soil
Pesticide residues, unexpected ingredients and packaging leachables
Impurities, E&Ls and contaminant screening
Extractables, leachables and degradation products
Botanical extracts, fragrance allergens and leachables
Explore more on the blog
Browse related discoveries and application notes by topic.
For Research Use Only. Not for use in diagnostic procedures.

