Prediction of retention indexes. II. Structure-retention index relationship on polar columns

AIRI: Predicting Retention Indices and Their Uncertainties Using Artificial Intelligence

The Kováts retention index (RI) is a quantity measured using gas chromatography and is commonly used in the identification of chemical structures. Creating libraries of observed RI values is a laborious task, so we explore the use of a deep neural network for predicting RI values from structure for standard semipolar columns. This network generated predictions with a mean absolute error of 15.1 and, in a quantification of the tail of the error distribution, a 95th percentile absolute error of 46.5. Because of the Artificial Intelligence Retention Indices (AIRI) network's accuracy, it was used to predict RI values for the NIST EI-MS spectral libraries. These RI values are used to improve chemical identification methods and the quality of the library. Estimating uncertainty is an important practical need when using prediction models. To quantify the uncertainty of our network for each individual prediction, we used the outputs of an ensemble of 8 networks to calculate a predicted standard deviation for each RI value prediction. This predicted standard deviation was corrected to follow the error between the observed and predicted RI values. The Z scores using these predicted standard deviations had a standard deviation of 1.52 and a 95th percentile absolute Z score corresponding to a mean RI value of 42.6.

Accurate prediction of isothermal gas chromatographic Kováts retention indices

We describe a freely available web server called Retention Index Predictor (RIpred) (https://ripred.ca) that rapidly and accurately predicts Gas Chromatographic Kováts Retention Indices (RI) using SMILES strings as chemical structure input. RIpred performs RI prediction for three different stationary phases (semi-standard non-polar (SSNP), standard non-polar (SNP), and standard polar (SP)) for both derivatized (trimethylsilyl (TMS) and tert‑butyldimethylsilyl (TBDMS) derivatized) and underivatized (base compound) forms of GC-amenable structures. RIpred was developed to address the need for freely available, fast, highly accurate RI predictions for a wide range of derivatized and underivatized chemicals for all common GC stationary phases. RIpred was trained using a Graph Neural Network (GNN) that used compound structures, their extracted features (mostly atom-level features) and the GC-RI data from the National Institute of Standards and Technology databases (NIST 17 and NIST 20). We curated this NIST 17 and NIST 20 GC-RI data, which is available for all three stationary phases, to create appropriate inputs (molecular graphs in this case) needed to enhance our model performance. The performance of different RIpred predictive models was evaluated using 10-fold cross validation (CV). The best performing RIpred models were identified and when tested on hold-out test sets from all stationary phases, achieved a Mean Absolute Error (MAE) of <73 RI units (SSNP: 16.5-29.5, SNP: 38.5-45.9, SP: 46.52-72.53). The Mean Absolute Percentage Error (MAPE) of these models were typically within 3% (SSNP: 0.78-1.62%, SNP: 1.87-2.88%, SP: 2.34-4.05%). When compared to the best performing model by Qu et al., 2021, RIpred performed similarly (MAE of 16.57 RI units [RIpred] vs. 16.84 RI units [Qu et al., 2021 predictor] for derivatized compounds). RIpred also includes ∼5 million predicted RI values for all GC-amenable compounds (∼57,000) in the Human Metabolome Database HMDB 5.0 (Wishart et al., 2022).

Yield and Composition of the Essential Oil of the Opopanax Genus in Turkey

The genus Opopanax W. Koch (Apiaceae) is represented by four species in Turkey. The composition of the essential oil of Opopanax genus members (Apiaceae) growing in Turkey was investigated in this study. GC-MS was used to analyze the composition of Opopanax essential oil samples that were taken from their natural environments. The Clevenger apparatus was used to hydrodistill the plant’s aerial parts, and the yields were determined to be between 0.2% v/w (for O. siifolius) and 0.4% (for O. hispidus, O. chironium, and O. persicus). The results and the chemical data provided some information and clues on the chemotaxonomy of the genus Opopanax. In this study, γ-elemene, butanoic acid octyl ester, and cylopropane were the main compounds identified in the essential oils of O. chironium, O. hispidus, and O. persicus. In particular, hexynyl n-valerate was most abundant in the essential oil of O. chironium, cyclopropane in that of O. hispidus, γ-elemene in that of O. persicus, and n-hexadecanoic acid/palmitic acid in that of O. siifolius. In a chemotaxonomic approach, the essential oil analysis of the Opopanax species revealed that these species conformed in a cluster analysis with their morphological classification. The constituents of the essential oils of all examined in the genus Opopanax were determined in this study, which is the most thorough one to date. This study provides new information about the composition of the essential oils of the investigated species.

Chemical Profile and Phytotoxic Action of Hibiscus trionum Essential Oil

The chemical profile and phytotoxic action of Hibiscus trionum essential oil (EO) was studied. In total 17 compounds were identified via GC/MS, representing 94.18 % of the entire oil, with phytol (40.37 %) being the dominant constituent. Bioassay revealed that the EO inhibited root elongation of Medicago sativa and Amaranthus retroflexus by 32.66 % and 61.86 % at 5 mg/mL, respectively; meanwhile, the major component phytol also exhibited significant phytotoxic activity, suppressing radical elongation of Pennisetum alopecuroides, M. sativa and A. retroflexus by 26.08 %, 27.55 % and 43.96 % at 1 mg/mL, respectively. The fact that the EO showed weaker activity than phytol implied that some constituents might trigger antagonistic action to decrease the oil's activity. Our study is the first on the chemical profile and phytotoxic effect of H. trionum EO.

Gas Chromatography Mass Spectrometry (GC-MS) Quantification of Metabolites in Stool Using 13C Labelled Compounds

It has become increasingly important to qualitatively and quantitatively assess the volatile metabolites in a range of bodily fluids for use in monitoring health. There has been relatively little work on the quantitative analysis of compounds, particularly with respect to the effects of ethnicity or geographic location. A novel method for the quantification of compounds in stool using ¹³C labelled compounds as internal standards is presented. Using thermal desorption gas chromatography mass spectrometry, stool samples from 38 healthy volunteers were analysed. The ¹³C labelled compounds, acetone, ethyl butanoate, ethanoic acid, butanoic acid, 3-methylbutanoic acid, and indole, were added as internal standards. This process mimics the solubility characteristics of the compounds and thus the method was able to quantify the compounds within the solid stool. In total, 15 compounds were quantified: Dimethyl sulphide (26–25,626 ng/g), acetone (442–3006 ng/g), ethyl butanoate (39–2468 ng/g), ethyl 2-methylbutanoate (0.3–180 ng/g), dimethyl disulphide (35–1303 ng/g), 1-octen-3-one (12 ng/g), dimethyl trisulphide (10–410 ng/g), 1-octen-3-ol (0.4–58 ng/g), ethanoic acid (672–12,963 ng/g), butanoic acid (2493–11,553 ng/g), 3-methylbutanoic acid (64–8262 ng/g), pentanoic acid (88–21,886 ng/g), indole (290–5477 ng/g), and 3-methyl indole (37–3483 ng/g). Moreover, by altering the pH of the stool to pH 13 in conjunction with the addition of ¹³C trimethylamine, the method was successful in detecting and quantifying trimethylamine for the first time in stool samples (range 40–5312 ng/g). Statistical analysis revealed that samples from U.K. origin had five significantly different compounds (ethyl butanoate, 1-octen-3-ol, ethanoic acid, butanoic acid, pentanoic acid, and indole) from those of South American origin. However, there were no significant differences between vegetarian and omnivore samples. These findings are supported by pre-existing literature evidence. Moreover, we have tentatively identified 12 compounds previously not reported as having been found in stool.

Alkylphenol retention indices

A novel type of retention indices for alkylphenols and related compounds are proposed. The alkylphenol retention indices (APRI) use para-substituted n-alkylphenols as reference series. APRI for para-n-alkylphenols are per definition equal to the number of carbon atoms in the alkyl substituent; the value for phenol is zero. Application of the APRI system with different types of derivatisation of the phenolic hydroxy group showed that the derivatisation has limited influence on these indices. Especially para-substituted alkylphenols gave APRI values that could be transferred with high accuracy from one type of derivative to another. By comparing results obtained with different gradients in temperature-programmed GC, it was also shown that APRI is less affected by chromatographic conditions than retention indices based on n-alkanes.

Standardized test mixture for the characterization of comprehensive two-dimensional gas chromatography columns: the Phillips mix

A novel column characterization test mixture is developed for use in comprehensive two-dimensional gas chromatography (GC x GC). This mixture has been named the "Phillips mix" in honor of the late professor John B. Phillips, the father of GC x GC. The mixture comprises a series of homologous compounds from structural groups that cover a volatility and polarity range that is similar to the Grob mix, and includes saturated hydrocarbons (alkanes), unsaturated hydrocarbons (alkenes and alkynes), carbonyls (ketones and aldehydes), primary alcohols, fatty acid methyl esters, alkyl ethers, carboxylic acids, aromatics, as well as other unique functional groups (such as amines, etc.). Similarly to the Grob mix in conventional one-dimensional GC, the Phillips mix can be used as a standardized test for performance characterization of GC x GC column sets. Unlike the Grob mix, however, the Phillips mix's most important use is as a practical guideline for column users. This paper addresses some qualitative aspects of the use of the Phillips mix through an investigation of the chromatographic fingerprints of two different GC x GC column combinations.

Prediction of retention indices. V. Influence of electronic effects and column polarity on retention index

The retention index increment for addition of a methylene group to an analyte molecule is shown for 1-halo-n-alkanes to be different from 100 i.u., a value that is customarily assigned according to the current convention in retention index prediction. In temperature-programmed gas chromatography using linearly interpolated retention index I, a linear regression equation, I=AZ+(GRF), with the number of atoms (Z) in the molecule as variable can describe the retention of 16 homologous series of organic compounds on non-polar and polar columns with characteristic A (linear regression coefficient) and (GRF) (group retention factor) values. A molecular model of retention on the basis of electron density and electron density distribution relative to that of n-alkane is proposed. This model brings out the inter- and intramolecular electronic effects in the analyte molecule and its dipole-dipole interaction with the stationary liquid phases, as variations in the A value. The (GRF) value varies with the connectivity ability of a functional group for extended conjugation, substitution, etc., but is most influenced by hydrogen bonding (H-bonding) with the stationary liquid phase. One can estimate the sequence of elution of a mixture of organic compounds from any two of the three parameters on the right-hand side of the above equation or retrieve the retention indexes of an entire homologous series from its A and (GRF) values. The fact that each analyte molecule has its own A value on different columns makes column difference (deltaI) compound-specific rather than column-specific, a departure from previous assumptions.

Carboxylic acids: prediction of retention data from chromatographic and electrophoretic behaviours

A review of the main results reached in the prediction of retention data of carboxylic acids, inferred by their chromatographic and electrophoretic behaviour, is presented. Attention has been focused on the main separation methods used in carboxylic acids analysis, that is ion-exclusion, anion-exchange, reversed-phase (RP) liquid chromatography and capillary electrophoresis. Papers proposing mechanistic models as well as chemometric and multilayer feed-forward neural network analysis of ion chromatography (IC) and RP chromatographic retention data were reviewed. Principal component analysis, PCA, sequential simplex method and simultaneous modelling of response surfaces through simple nonlinear models (not related to equilibria involved in retention) have been considered. Computer simulations for the prediction of retention data have also been discussed. A quick overlook on the prediction of capacity factors of analytes by less common determination methods such as thin-layer, gas chromatography and supercritical fluid chromatography has also been done.

Retrieval of structure information from retention index

The chromatographic identity of a compound can be determined by four parameters, namely, I, A, Z and (GRF). These are interrelated in a linear regression equation, given in the paper as Eq. 8. The retrieval of structural information from retention data requires the introduction of a new meaning to the Kováts retention index, the use of column difference (delta I) to characterize functional groups, the redefinition of the role of electronegative oxygen and nitrogen atoms, and the division of retention index (I) into contributions from atoms and from functional groups. The separation of retention index (I) into molecular and interaction contributions is a necessary condition for retention index prediction from structure and also for structure information retrieval from retention data. According to Eq. 8 the retention index is uniquely determined by three parameters, namely A, Z and (GRF). For prediction of retention index, the A value is assigned a value of 100 index units (i.u.), the Z value is obtained directly from the compound, and the (GRF) value is pre-calibrated. In Eq. 10, the m and n values represent the pre-calibrated terms for a quantitative structure-retention index relationship. These terms account for the positive and negative retention contributions from polar and polarizable atom groups. All atom groups that are different from methylene and methyl groups will interact with the stationary phase and contribute to retention. The m and n values for various functional, polar and polarizable atom groups and their column differences (delta I values) are the results of interactions between the solute and the stationary phase and are structure dependent. The interaction increases with increasing polarities of the solute and the stationary phase. The column difference not only reflects the strength of the interaction, but is also characteristic of the functional and polarizable groups. The retrieval of structural information from retention data is equivalent to obtaining Z and (GRF) values from known I and delta I values, which is straightforward for monofunctional compounds. For multi-functional compounds, additional data will be needed for retrieval of structural information. These can be obtained from derivatization of the unknown compound, from its chemical reactions with other reagents, from GC-MS analysis and from structure match using internal or external standards. The additional data required will depend upon the complexity of the unknown structure. This approach demonstrates that a system can be devised to utilize GC retention characteristics uniquely for structure elucidation.

Prediction of retention indexes. IV. Chain branching in alkylbenzene isomers with C10-13 alkyl chains identified in a scintillator solvent

Twenty solvent components in a commercial scintillator were identified by chromatography on polar and non-polar columns and by gas chromatography-mass spectrometry (GC-MS) as isomeric 1-(alkyl)m(alkyl)nbenzenes with formulae C16H26, C17H28, C18H30 and C19H32. These isomers occur in four clusters of chromatographic peaks representing ca. 6, 44, 34 and 16% of the total solvent mass. The retention indexes of the isomers are influenced by the lengths of the alkyl chains in the molecule, and their polarity and polarizability can affect the column difference, which is the difference between retention indexes on polar and non-polar columns. 1-Methylalkylbenzenes have higher retention indexes and larger column differences than the evenly distributed isomers, such as 1-butylhexyl-1-pentylhexyl, 1-pentylheptyl- and 1-pentyloctylbenzene. The results demonstrate the effect of structural symmetry on the retention indexes of the isomers. This study shows that the ability to relate GC data and column differences to structures can facilitate the interpretation of GC-MS data in the structure identification of isomers.

Prediction of retention indexes. III. Silylated derivatives of polar compounds

Polar compounds containing hydroxyl, amino and carboxyl groups, singly or in combination, can be chromatographed after the polar functional groups are silylated. The silylated derivatives of acids, alcohols, amines, diols, amino alcohols, amino acids are shown to behave chromatographically as hydrocarbons, and their retention indexes can be readily predicted from their base values. The column difference, namely, the difference between the retention indexes of the analyte on polar and non-polar columns is minimal for the silylated derivatives in comparison to that observed for the underivatized analytes. This minimal column difference is attributed to the hydrocarbon-like chromatographic characteristics of the silylated derivatives. The retention indexes of the silyl derivatives appear to correlate with the atom number Z of the analyte.