FRACTAL: A Ratiometric-based Fingerprinting Framework for Robust Spectral Classification and Ratio-based Spectroscopic Marker Discovery
Spectral classification is conventionally performed on band intensities, yet ratios between bands have long served as diagnostic indices in vibrational spectroscopy, from lipid-to-protein markers of membrane composition to amide ratios reporting protein secondary structure. Their use, however, has remained confined to a handful of expert-selected pairs, leaving most of the ratio information contained in a spectrum unexamined. Here I present FRACTAL (Full Ratio Combinatorial Analysis for Classification and Analytical Interpretation), which computes every pairwise log-intensity ratio among detected peaks to form a complete ratiometric fingerprint, grounded in compositional data analysis. Each component is a contrast between two assigned bands and is invariant to the multiplicative factors that scale spectral intensities. Applied to serum Raman spectra from 103 donors (53 with COVID-19, 50 healthy controls), FRACTAL screened all 91 ratios among 14 reproducible peaks and identified 44 discriminant contrasts. A ten-ratio fingerprint reached a mean internal AUC of 0.93–0.96 across six classifier architectures spanning linear, kernel-based and ensemble methods, exceeding the corresponding peak intensities for every classifier and are within 0.04 AUC of the full 457-point spectrum. It generalized to a withheld 20-donor cohort (AUC 0.94–0.96) with the narrowest spread across classifiers of any representation tested, whereas the single most discriminant ratio, despite performing well internally, detected only half of the withheld COVID-19 donors. The top-ranked contrasts include a well-known lipid-to-protein ratio (1451/1662 cm⁻¹) and are consistent with the altered protein composition and aromatic amino acid metabolism reported in COVID-19. FRACTAL establishes exhaustive ratiometric fingerprinting as a compact, robust and interpretable representation for spectral classification and spectroscopic marker discovery.
Authors
- Lamyaa M. Almehmadi (ORCID: https://orcid.org/0000-0002-2572-9733)
Institutions
- Massachusetts Institute of Technology (US)
Publication Details
- Journal
- ChemRxiv
- Published
- 2026-10-27
- DOI
- https://doi.org/10.26434/chemrxiv.15010325/v2
- Primary Topic
- Spectroscopy Techniques in Biomedical and Chemical Research
- Type
- preprint