Jones — PFAS Analytical Chemistry & blueFish
Emeritus domain expert providing the product specification and validation rubric for blueFish through decades of analytical chemistry expertise.
Status: Active — emeritus consulting confirmed, technical resources received Domain: PFAS analytical chemistry, mass spectrometry, environmental monitoring Product: blueFish (sovereign PFAS analytical chemistry ETL) Springs Fed: wetSpring Track 2
The Collaboration
This is not a typical gen5 collaboration (PI + grant). It is an emeritus domain expert providing the product specification and validation rubric for blueFish through accumulated institutional knowledge — 30+ years of analytical chemistry domain expertise.
What Domain Expertise Revealed
EPA Method 1633A Gap
Most labs (including commercial) do not actually perform all QC checks — particularly ion abundance ratios. No published data exists to evaluate QC compliance beyond single-letter qualifiers in lab reports. blueFish implements full 1633A QC including ion abundance ratios — the gap commercial labs are not filling.
Open-Source Tool Assessment
| Tool | Assessment |
|---|---|
| MZmine | Unacceptable failure rates |
| MS-DIAL | Unacceptable failure rates |
| XCMS | Unacceptable failure rates |
| Progenesis QI | Good graphical evaluation, now obsolete |
No open-source tool gives acceptable failure rates AND graphical evaluation of feature detection / measurement quality. This is blueFish’s target.
MRM Architecture
Most PFAS analyses are targeted MRM (Multiple Reaction Monitoring). Single sample produces ~100 chromatograms (2 per analyte + internal standards). Current lab workflow uses manual spreadsheet consolidation. blueFish implements template-driven MRM batch processing with automated peak integration.
blueFish Product Specification (from Domain Data)
| Feature | Source | Evidence |
|---|---|---|
| Full EPA 1633A QC (including ion abundance ratios) | Domain expertise | “Most labs don’t actually perform all QC checks” |
| LOQ boundary handling | Domain expertise | State labs report below-LOQ as equal to LOQ |
| Visual QC for feature detection | Domain expertise | Progenesis QI was good at this but is now dead |
| Low failure rate peak detection | Domain expertise | MZmine, MS-DIAL, XCMS all have unacceptable failure rates |
| Sovereign mzML ingestion | Domain expertise | Chromeleon to msConvert to mzML is the standard path |
| NIST reference validation | Domain expertise | Untargeted PFAS reference spectra as validation baseline |
| MANA SODA benchmarking | Domain expertise | Cross-validation consortium datasets for quality review |
| Template-driven MRM batch processing | Domain expertise | ~100 chromatograms/sample, tedious manual setup |
New Spring Validation Targets
| Target | Spring Check |
|---|---|
| EPA 1633A ion abundance ratio QC | Full method compliance checks |
| LOQ boundary validation | Below-LOQ detection and reporting accuracy |
| NIST reference spectra matching | Untargeted PFAS validation suite |
| MANA SODA cross-validation parity | Multi-tool benchmark comparison |
| MRM peak integration accuracy | 100-chromatogram batch processing |
These are validation targets the ecosystem could never have generated internally — they come from decades of analytical chemistry domain expertise.
New Public Data Systems
| System | Purpose | Status |
|---|---|---|
| NIST PFAS Reference Data | Untargeted PFAS validation baseline | Planned |
| EPA Method 1633A | Full QC compliance specification | Planned |
| MANA SODA Benchmark Datasets | Cross-validation consortium | Planned |
The gen5 Pattern
Domain expertise (decades of analytical chemistry)
-> Technical data (NIST, EPA 1633A, MANA SODA, MRM architecture)
-> blueFish product specification
-> wetSpring Track 2 new validation targets
-> blueFish implementation (sovereign PFAS ETL)
-> Architecture and QC compliance review
-> blueFish available to any PFAS lab (AGPL-3.0)
-> Domain knowledge becomes embodied in open tooling
Retirement from the university is not retirement from the science. Decades of analytical chemistry expertise — what tools fail, what QC labs skip, where the data quality gaps are — is now the specification for an open tool that outlives any single institution.