Computational chromatography: A machine learning strategy for demixing individual chemical components in complex mixtures

Abstract

Surface-enhanced Raman spectroscopy (SERS) holds exceptional promise as a streamlined chemical detection strategy for biological and environmental contaminants compared with current laboratory methods. Priority pollutants such as polycyclic aromatic hydrocarbons (PAHs), detectable in water and soil worldwide and known to induce multiple adverse health effects upon human exposure, are typically found in multicomponent mixtures. By combining the molecular fingerprinting capabilities of SERS with the signal separation and detection capabilities of machine learning (ML), we examine whether individual PAHs can be identified through an analysis of the SERS spectra of multicomponent PAH mixtures. We have developed an unsupervised ML method we call Characteristic Peak Extraction, a dimensionality reduction algorithm that extracts characteristic SERS peaks based on counts of detected peaks of the mixture. By analyzing the SERS spectra of two-component and four-component PAH mixtures where the concentration ratios of the various components vary, this algorithm is able to extract the spectra of each unknown component in the mixture of unknowns, which is then subsequently identified against a SERS spectral library of PAHs. Combining the molecular fingerprinting capabilities of SERS with the signal separation and detection capabilities of ML, this effort is a step toward the computational demixing of unknown chemical components occurring in complex multicomponent mixtures.

Speciation of Nanocatalysts Using X-ray Absorption Spectroscopy Assisted by Machine Learning

“…We note that VAEs are only one example of the deep generative models that have an explicit model of probabilistic distribution learning built in . While the use of invertible embeddings is still in infancy in XAFS, other fields of chemistry and catalysis in general have adapted these methods for a range of applications, from denoising, demixing, and classification to generating potential solutions to ill-posed problems . Recent revolution of generative models has been very successful in utilizing these invertible embeddings for “inverting” novel structures of small molecules with targeted properties, and efforts are underway to expand the application to crystalline materials .…”

Identifying Surface-Enhanced Raman Spectra with a Raman Library Using Machine Learning

“…The direction reported here is essential for advancing the utility of SERS as a research tool and as a sensing modality that yields chemically specific information. We foresee substantial opportunities and benefits of further testing, improving and refining this algorithm and applying it to other surface-enhanced spectroscopies, such as surface-enhanced IR absorption spectroscopy, − and other procedures that require spectral recognition, such as identifying chemicals from mixtures across different types of spectroscopies …”

“…Then, CaPSim is then calculated in the following procedure: Given a query q and a library scriptL = { R 1 , ··· , R N } , we use an algorithm scriptA to extract the potential CP locations from each molecule’s spectra, i.e., scriptK j = scriptA ( R j ) , ∀ j ∈[ N ]. After obtaining scriptK j , we perform a spectral max-pooling over each spectrum r i , j in R j , ∀ i ∈[ n j ], and obtain the compressed spectrum r̃ i , j = false[ max k ∈ K 1 , j r i , j , k , max k ∈ K 2 , j r i , j , k , ··· , max k ∈ K m j , j r i , j , k false] T . In other words, for each set of potential CP indices K l , j , ∀ l ∈[ m j ], we take the maximum intensity in this specific area of the spectrum r ...…”

“…After obtaining scriptK j , we perform a spectral max-pooling over each spectrum r i , j in R j , ∀ i ∈[ n j ], and obtain the compressed spectrum r̃ i , j = false[ max k ∈ K 1 , j r i , j , k , max k ∈ K 2 , j r i , j , k , ··· , max k ∈ K m j , j r i , j , k false] T . In other words, for each set of potential CP indices K l , j , ∀ l ∈[ m j ], we take the maximum intensity in this specific area of the spectrum r i , j .…”

“…Recently, an unsupervised ML strategy for demixing a complex mixture of SERS spectra of various molecules was proposed, 40 which relies on Characteristic Peak Extraction (CaPE). CaPE is a spectral feature extraction algorithm that yields characteristic SERS peaks based on counts of detected peaks from spectra of a mixture, taking relatively small frequency shifts into consideration.…”