Multivariate Analysis of Mixed Lipid Aggregate Phase Transitions Monitored Using Raman Spectroscopy

Abstract

The phase behavior of aqueous 1,2-dimyristoyl-sn-glycero-3-phosphorylcholine (DMPC)/1,2-dihexanoyl-sn-glycero-3-phosphocholine (DHPC) mixtures between 8.0 ℃ and 41.0 ℃ were monitored using Raman spectroscopy. Temperature-dependent Raman matrices were assembled from series of spectra and subjected to multivariate analysis. The consensus of pseudo-rank estimation results is that seven to eight components account for the temperature-dependent changes observed in the spectra. The spectra and temperature response profiles of the mixture components were resolved by applying a variant of the non-negative matrix factorization (NMF) algorithm described by Lee and Seung (1999). The rotational ambiguity of the data matrix was reduced by augmenting the original temperature-dependent spectral matrix with its cumulative counterpart, i.e., the matrix formed by successive integration of the spectra across the temperature index (columns). Successive rounds of constrained NMF were used to isolate component spectra from a significant fluorescence background. Five major components exhibiting varying degrees of gel and liquid crystalline lipid character were resolved. Hydrogen-bonded water networks exhibiting varying degrees of organization are associated with the lipid components. Spectral parameters were computed to compare the chain conformation, packing, and hydration indicated by the resolved spectra. Based on spectral features and relative amounts of the components observed, four components reflect long chain lipid response. The fifth component could reflect the response of the short chain lipid, DHPC, but there were no definitive spectral features confirming this assignment. A minor component of uncertain assignment that exhibits a striking response to the DMPC pre-transition and chain melting transition also was recovered. While none of the spectra resolved exhibit features unequivocally attributable to a specific aggregate morphology or step in the gelation process, the results are consistent with the evolution of mixed phase bicelles (nanodisks) and small amounts of worm-like DMPC/DHPC aggregates, and perhaps DHPC micelles, at low temperature to suspensions of branched and entangled worm-like aggregates above the DMPC gel phase transition and perforated multi-lamellar aggregates at high temperature.

OH-Stretch Raman Multivariate Curve Resolution Spectroscopy of HOD/H2O Mixtures

“…One of the two spectral components is equivalent to a pure solvent spectrum, whereas the other contains SC spectral features . However, no assumptions regarding the spectral shape are required. , Raman MCR has been used to study solutions of acetonitrile in water, π-hydrogen bond formation between liquid water and benzene, dangling hydroxyl groups of water around benzyl alcohol, the hydration shell structure of CO 2 and hydrogen bonding between CO 2 and water, hydrophobic hydration shells of n -alcohols, phase transitions in mixed lipid aggregates, aqueous proton solvation, the affinity of ions for hydrophobic hydration shells, and for solutions of benzene and pyridine in water . Raman MCR has also been used to decompose temperature-dependent Raman spectra of liquid H 2 O and D 2 O into linear combinations of two spectral components with temperature-independent shapes but temperature-dependent intensities, corresponding to water molecules in different local tetrahedral environments. , …”

Rise of Raman spectroscopy in neurosurgery: a review

“…Deep neural network architectures such as convolutional neural networks (CNN) have shown great promises in biospectroscopy, with the additional benefit of being less dependent on spectral preprocessing. [163][164][165][166][167][168][169] Data requirements to train and optimize such models are high because of the millions of parameters they contain; [170][171][172] nevertheless, open access to large Raman datasets and strategies such as transfer learning and novel data augmentation methods (such as the simulation of Raman spectra for DRS analysis 173 ) will make their adoption possible for biomedical applications in the near future. 174…”

Wavelength calibration with deep-learning using long short-term memory architectures

“…9 By training on extensive training sets deep learning has already been established as a useful technology for processing spectra. [21][22][23][24][25] Of particular interest in this paper is the long-term short-term memory (LSTM) network. An LSTM is a sub class of Recurrent Neural Network (RNN), where the network is trained by looping through sequences of data as functions of time.…”