LC Retention Time Research Articles

Phytosterols in human serum as measured using a liquid chromatography tandem mass spectrometry

Phytosterols are lipophilic compounds found in plants with structural similarity to mammalian cholesterol. They cannot be endogenously produced by mammals and therefore always originate from diet. There has been increased interest in dietary phytosterols over the last few decades due to their association with a variety of beneficial health effects including low-density lipoprotein cholesterol lowering, anti-inflammatory and anti-cancerous effects. They are proposed as potential moderators for diseases associated with the central nervous system where cholesterol homeostasis is found to be imperative (multiple sclerosis, dementia, etc.) due to their ability to reach the brain. Here we utilised an enzyme-assisted derivatisation for sterol analysis (EADSA) in combination with a liquid chromatography tandem mass spectrometry (LC-MSn) to characterise phytosterol content in human serum. As little as 100 fg of plant sterol was injected on a reversed phase LC column. The method allows semi-quantitative measurements of phytosterols and their derivatives simultaneously with measurement of cholesterol metabolites. The identification of phytosterols in human serum was based on comparison of their LC retention times and MS2, MS3 spectra with a library of authentic standards. Free campesterol serum concentration was in the range from 0.30-4.10 µg/mL, β-sitosterol 0.16-3.37 µg/mL and fucosterol was at lowest concentration range from 0.05-0.38 µg/mL in ten individuals. This analytical methodology could be applied to the analysis of other biological fluids and tissues.

MSBooster: improving peptide identification rates using deep learning-based features

Peptide identification in liquid chromatography-tandem mass spectrometry (LC-MS/MS) experiments relies on computational algorithms for matching acquired MS/MS spectra against sequences of candidate peptides using database search tools, such as MSFragger. Here, we present a new tool, MSBooster, for rescoring peptide-to-spectrum matches using additional features incorporating deep learning-based predictions of peptide properties, such as LC retention time, ion mobility, and MS/MS spectra. We demonstrate the utility of MSBooster, in tandem with MSFragger and Percolator, in several different workflows, including nonspecific searches (immunopeptidomics), direct identification of peptides from data independent acquisition data, single-cell proteomics, and data generated on an ion mobility separation-enabled timsTOF MS platform. MSBooster is fast, robust, and fully integrated into the widely used FragPipe computational platform.

Complementary methods for structural assignment of isomeric candidate structures in non-target liquid chromatography ion mobility high-resolution mass spectrometric analysis

Non-target screening with LC/IMS/HRMS is increasingly employed for detecting and identifying the structure of potentially hazardous chemicals in the environment and food. Structural assignment relies on a combination of multidimensional instrumental methods and computational methods. The candidate structures are often isomeric, and unfortunately, assigning the correct structure among a number of isomeric candidate structures still is a key challenge both instrumentally and computationally. While practicing non-target screening, it is usually impossible to evaluate separately the limitations arising from (1) the inability of LC/IMS/HRMS to resolve the isomeric candidate structures and (2) the uncertainty of in silico methods in predicting the analytical information of isomeric candidate structures due to the lack of analytical standards for all candidate structures. Here we evaluate the feasibility of structural assignment of isomeric candidate structures based on in silico–predicted retention time and database collision cross-section (CCS) values as well as based on matching the empirical analytical properties of the detected feature with those of the analytical standards. For this, we investigated 14 candidate structures corresponding to five features detected with LC/HRMS in a spiked surface water sample. Considering the predicted retention times and database CCS values with the accompanying uncertainty, only one of the isomeric candidate structures could be deemed as unlikely; therefore, the annotation of the LC/IMS/HRMS features remained ambiguous. To further investigate if unequivocal annotation is possible via analytical standards, the reversed-phase LC retention times and low- and high-resolution ion mobility spectrometry separation, as well as high-resolution MS2 spectra of analytical standards were studied. Reversed-phase LC separated the highest number of candidate structures while low-resolution ion mobility and high-resolution MS2 spectra provided little means for pinpointing the correct structure among the isomeric candidate structures even if analytical standards were available for comparison. Furthermore, the question arises which prediction accuracy is required from the in silico methods to par the analytical separation. Based on the experimental data of the isomeric candidate structures studied here and previously published in the literature (516 retention time and 569 CCS values), we estimate that to reduce the candidate list by 95% of the structures, the confidence interval of the predicted retention times would need to decrease to below 0.05 min for a 15-min gradient while that of CCS values would need to decrease to 0.15%. Hereby, we set a clear goal to the in silico methods for retention time and CCS prediction.Graphical abstract

Machine learning liquid chromatography retention time prediction model augments the dansylation strategy for metabolite analysis of urine samples

Herein, a standalone software equipped with a graphic user interface (GUI) is developed to predict liquid chromatography mass spectrometry (LC–MS) retention times (RTs) of dansylated metabolites. Dansylation metabolomics strategy developed by Li et al. narrows down a vast chemical space of metabolites into the metabolites containing amines and phenolic hydroxyls. Combined with differential isotope labeling, e.g., 12C-reagent labeled individual samples spiked with a 13C-reagent labeled reference or pooled sample, LC–MS analysis of the dansylated samples enables accurate relative quantification of all labeled metabolites. Herein, the LC–RTs for dansylated metabolites are predicted using an artificial neural network (ANN) machine-learning model. For the ANN modeling, 315 dansylated urine metabolites obtained from the DnsID database are used. The ANN LC–RT prediction model was reliable, with a mean absolute deviation of 0.74 min for the 30 min LC run. In the RT model, a deviation of more than 2 min was observed in only 3.2% of the total 315 metabolites, while a deviation of 1.5 min or more was observed in 11% of the metabolites. Furthermore, it was found that the LC–RT prediction was also reliable even for metabolites containing both amine and phenolic functional groups that can undergo dansylation on either one of the two functional groups, resulting in the generation of two isomeric forms. This RT-prediction model is embedded into a user-friendly GUI and can be used for identifying nontargeted dansylated metabolites with unknown RTs, along with accurate mass measurements. Furthermore, it is demonstrated that the developed software can help identify metabolites from a urine sample of an anonymous healthy pregnant woman.

Use of Liquid Chromatography--Tandem Mass Spectrometry to Quantify and Confirm the Fentanyl Metabolite N-[1-(2-Phenethy-4-Piperidinyl)] Maloanilinic Acid in Equine Urine for Doping Control.

Fentanyl, a powerful synthetic mu opioid receptor agonist, is banned in equine sports by the Association of Racing Commissioners International and the Fédération Équestre Internationale. The presence of fentanyl in equine blood has been confirmed during routine post-race screening for doping substances in the authors' laboratory. While fentanyl can be detected and confirmed in blood, it is rapidly metabolized, and screening for the metabolite N-[1-(2-phenethy-4-piperidinyl)] maloanilinic acid (PMA) in equine urine is expected to allow for a longer detection time. In this study, a quantitative and confirmatory liquid chromatography--tandem mass spectrometry (LC-MS-MS) method was developed for PMA analysis in equine urine. PMA was extracted by solid phase extraction, separated on a C18 column and detected using a triple quadrupole mass spectrometer. The mass spectrometer was operated in positive-ion mode, and multiple reaction monitoring was used to monitor product ions m/z 188, m/z 281 and m/z 323. The method was validated for extraction recovery, matrix effect, specificity, sensitivity, precision and accuracy, carryover and processed sample stability according to the guidelines of the US Food and Drug Administration for bioanalysis. The limits of detection and quantification were 5 and 10 pg/mL, respectively. Linearity was obtained over the concentration range of 10-10,000 pg/mL. To confirm PMA in equine urine, LC retention time, diagnostic product ions (m/z 188, m/z 281 and m/z 323) and product ion ratio were used as the criteria. The lowest concentration for confirmatory analysis was validated at 50 pg/mL. The method was applied to measure the PMA concentrations in equine urine following intravenous administration of fentanyl to a research horse and has confirmed the presence of PMA in post-race urine samples. This method is a valuable addition to the arsenal of equine doping control methods to combat illegal doping and protect racehorse health.

Separation and identification of a complex flurbiprofen-polyethylene glycol mono- and diester mixture via a hyphenated HPLC-DAD-HRMS/SPE NMR system

We report the use of a hyphenated HPLC-DAD(diode array detector)-HRMS/SPE NMR system for the separation and isolation of a complex mixture of esters, containing substances with very similar LC retention times. The literature known mono- and diesters of the drug Flurbiprofen and polyethylene glycol, which form a large number of substances with varying chain lengths, were chosen for this study. We demonstrate the use of this hyphenated system to quickly and effectively isolate sixteen of these highly similar individual esters in an automated fashion, demonstrating its applicability in standard pharmaceutical analysis and quality control of drugs. Both, synthetic solutions of these esters and extracts from Flurbiprofen lozenges were used for this purpose. By the sole use of this system, the individual compounds were isolated and UV, HRMS and 1D and 2D NMR data could be collected, enabling the identification and differentiation of the individual esters.

A Data Set of Ion Mobility Collision Cross Sections and Liquid Chromatography Retention Times from 71 Pyridylaminated N-Linked Oligosaccharides

Determination of the glycan structure is an essential step in understanding structure–function relationships of glycans and glycoconjugates including biopharmaceuticals. Mass spectrometry, because of its high sensitivity and mass resolution, is an excellent means of analyzing glycan structures. We previously proposed a method for rapid and precise identification of N-glycan structures by ultraperformance liquid chromatography-connected ion mobility mass spectrometry (UPLC/IM-MS). To substantiate this methodology, we here examine 71 pyridylaminated (PA-) N-linked oligosaccharides including isomeric pairs. A data set on collision drift times, retention times, and molecular mass was collected for these PA-oligosaccharides. For standardization of the observables, LC retention times were normalized into glucose units (GU) using pyridylaminated α-1,6-linked glucose oligomers as reference, and drift times in IM-MS were converted into collision cross sections (CCS). To evaluate the CCS value of each PA-oligosaccharide, we introduced a CCS index which is defined as a CCS ratio of a target PA-glycan to the putative standard PA-glucose oligomer of the same m/z. We propose a strategy for practical structural analysis of N-linked glycans based on the database of m/z, CCS index, and normalized retention time (GU).

Sterol profiling of Leishmania parasites using a new HPLC-tandem mass spectrometry-based method and antifungal azoles as chemical probes reveals a key intermediate sterol that supports a branched ergosterol biosynthetic pathway

Human leishmaniasis is an infectious disease caused by Leishmania protozoan parasites. Current chemotherapeutic options against the deadly disease have significant limitations. The ergosterol biosynthetic pathway has been identified as a drug target in Leishmania. However, remarkable differences in the efficacy of antifungal azoles that inhibit ergosterol biosynthesis have been reported for the treatment of leishmaniasis. To better understand the sterol biosynthetic pathway in Leishmania and elucidate the mechanism underlying the differential efficacy of antifungal azoles, we developed a new LC-MS/MS method to study sterol profiles in promastigotes of three Leishmania species, including two L. donovani, one L. major and one L. tarentolae strains. A combination of distinct precursor ion masses and LC retention times allowed for specific detection of sixteen intermediate sterols between lanosterol and ergosterol using the newly developed LC-MS/MS method. Although both posaconazole and fluconazole are known inhibitors of fungal lanosterol 14α-demethylase (CYP51), only posaconazole led to a substantial accumulation of lanosterol in azole-treated L. donovani promastigotes. Furthermore, a key intermediate sterol accumulated by 40- and 7-fold when these parasites were treated with posaconazole and fluconazole, respectively, which was determined as 4α,14α-dimethylzymosterol by high resolution mass spectrometry and NMR spectroscopy. The identification of 4α,14α-dimethylzymosterol supports a branched ergosterol biosynthetic pathway in Leishmania, where lanosterol C4- and C14-demethylation reactions occur in parallel rather than sequentially. Our results suggest that selective inhibition of leishmanial CYP51 is insufficient to effectively prevent parasite growth and dual inhibitors of both CYP51 and the unknown sterol C4-demethylase may be required for optimal antiparasitic effect.

Distinguishing Histidine Tautomers in Proteins Using Covalent Labeling-Mass Spectrometry.

In this work, we use diethylpyrocarbonate (DEPC)-based covalent labeling together with LC-MS/MS analysis to distinguish the two sidechain tautomers of histidine residues in peptides and proteins. From labeling experiments on model peptides, we demonstrate that DEPC reacts equally with both tautomeric forms to produce chemically different products with distinct dissociation patterns and LC retention times, allowing the ratios of the two tautomers to be determined in peptides and proteins. Upon measuring the tautomer ratios of several histidine residues in myoglobin, we find good agreement with previous 2D NMR data on this protein. Because our DEPC labeling/MS approach is simpler, faster, and more precise than 2D NMR, our method will be a valuable way to determine how protein structure enforces histidine sidechain tautomerization. Because the tautomeric state of histidine residues is often important for protein structure and function, the ability of DEPC labeling/MS to distinguish histidine tautomers should equip researchers with a tool to understand the histidine residue structure and function more deeply in proteins.

A mathematical method for calibrating the signal drift in liquid chromatography - mass spectrometry analysis

Liquid chromatography coupled with high-resolution mass spectrometry (LC-HRMS) has become the most versatile analytical tool for profiling small-molecule compounds and increasingly been applied in many fields. Nevertheless, LC-MS based quantification still face some challenges, such as signal drift in LC-MS, which may affect the validity of the obtained data and lead to misinterpretation of biological results. Here, we established a calibration method known as “RIM” to compensate the signal drift of LC-MS. To this end, a mixture of d4-2-dimethylaminoethylamine (d4-DMED)-coded normal fatty acids (C5–C23) was used as calibrants to construct RIM calibration. With the addition of calibrants, not only the MS signal drift, but also the mass accuracy and LC retention time can be calibrated, thereby improving the reliability of quantitative data. The effectiveness of RIM was carefully validated using a human serum extract spiked with 34 standards and then RIM was applied for rat brain untargeted metabolome research. In addition, to expand the functionality and flexibility of RIM for data handling, we generated a MATLAB-based RIM program, which implements the above concepts and allows automatic data process. Taken together, the proposed RIM method has potential application in large-scale quantitative study of complex samples.

Isolation, characterization and semi-synthesis of natural products dimeric amide alkaloids

Isolation, characterization and semi-synthesis of natural products dimeric amide alkaloids

Elucidation of Gut Microbiota-Associated Lipids Using LC-MS/MS and 16S rRNA Sequence Analyses.

SummaryHost-microbiota interactions create a unique metabolic milieu that modulates intestinal environments. Integration of 16S ribosomal RNA (rRNA) sequences and mass spectrometry (MS)-based lipidomics has a great potential to reveal the relationship between bacterial composition and the complex metabolic network in the gut. In this study, we conducted untargeted lipidomics followed by a feature-based molecular MS/MS spectral networking to characterize gut bacteria-dependent lipid subclasses in mice. An estimated 24.8% of lipid molecules in feces were microbiota-dependent, as judged by > 10-fold decrease in antibiotic-treated mice. Among these, there was a series of unique and microbiota-related lipid structures, including acyl alpha-hydroxyl fatty acid (AAHFA) that was newly identified in this study. Based on the integrated analysis of 985 lipid profiles and 16S rRNA sequence data providing 2,494 operational taxonomic units, we could successfully predict the bacterial species responsible for the biosynthesis of these unique lipids, including AAHFA.

Structural elucidation of amino amide-type local anesthetic drugs and their main metabolites in urine by LC-MS after derivatization and its application for differentiation between positional isomers of prilocaine.

The demand for clinical toxicology analytical methods for identifying drugs of abuse and medicinal drugs is steadily increasing. Structural elucidation of amino amide-type local anesthetic drugs and their main metabolites by GC-EI-MS and LC-ESI-MS/MS is of great analytical challenge. These compounds exhibit only/mostly fragments/product ions representing the amine-containing residue, while the aromatic amide moiety remains unidentified. This task becomes even more complicated when discrimination between positional isomers of such compounds is required. Here, we report the development of a derivatization procedure for the differentiation and structural elucidation of a mixture of local anesthetic drugs and their metabolites that possess tertiary and secondary amines in water and urine. A method based on two sequential "in-vial" instantaneous derivatization processes at ambient temperature followed by LC-ESI-MS/MS analysis was developed. 2,2,2-Trichloro-1,1-dimethylethyl chloroformate (TCDMECF) was utilized to selectively convert the secondary amines into their carbamate derivatives, followed by hydrogen peroxide addition to produce the corresponding tertiary amine oxides. The resulting derivatives exhibited rich fragmentation patterns, enabling improved structural elucidation of the original compounds. The developed method was successfully applied to the differentiation and structural elucidation of prilocaine and its four positional isomers, which all possess similar GC and LC retention times and four of them exhibit almost identical EI-MS and ESI-MS/MS spectra, enabling their structural elucidation in a single LC-ESI-MS/MS analysis. The developed technique is fast and simple and enables discrimination between isomers based on different diagnostic ions/fragmentation patterns.

A Unique Combination of Two Different Quorum Sensing Systems in the β-Rhizobium Cupriavidus taiwanensis.

Cupriavidus taiwanensis LMG19424, a β-rhizobial symbiont of Mimosa pudica, harbors phc and tqs quorum sensing (QS), which are the homologous cell-cell communication systems previously identified from the plant pathogen Ralstonia solanacearum and the human pathogen Vibrio cholerae, respectively. However, there has been no experimental evidence reported that these QS systems function in C. taiwanensis LMG19424. We identified (R)-methyl 3-hydroxymyristate (3-OH MAME) and (S)-3-hydroxypentadecan-4-one (C15-AHK) as phc and tqs QS signals, respectively, and characterized these QS systems. The expression of the signal synthase gene phcB and tqsA in E. coli BL21(DE3) resulted in the high production of 3-OH MAME and C15-AHK, respectively. Their structures were elucidated by comparison of EI-MS data and GC/chiral LC retention times with synthetic standards. The deletion of phcB reduced cell motility and increased biofilm formation, and the double deletion of phcB/tqsA caused the accumulation of the metal chelator coproporphyrin III in its mutant culture. Although the deletion of phcB and tqsA slightly reduced its ability to nodulate on aseptically grown seedlings of M. pudica, there was no significant difference in nodule formation between LMG19424 and its QS mutants when commercial soils were used. Taken together, this is the first example of the simultaneous production of 3-OH MAME/C15-AHK as QS signals in a bacterial species, and the importance of the phc/tqs QS systems in the saprophytic stage of C. taiwanensis LMG19424 is suggested.

JUMPm: A Tool for Large-Scale Identification of Metabolites in Untargeted Metabolomics.

Metabolomics is increasingly important for biomedical research, but large-scale metabolite identification in untargeted metabolomics is still challenging. Here, we present Jumbo Mass spectrometry-based Program of Metabolomics (JUMPm) software, a streamlined software tool for identifying potential metabolite formulas and structures in mass spectrometry. During database search, the false discovery rate is evaluated by a target-decoy strategy, where the decoys are produced by breaking the octet rule of chemistry. We illustrated the utility of JUMPm by detecting metabolite formulas and structures from liquid chromatography coupled tandem mass spectrometry (LC-MS/MS) analyses of unlabeled and stable-isotope labeled yeast samples. We also benchmarked the performance of JUMPm by analyzing a mixed sample from a commercially available metabolite library in both hydrophilic and hydrophobic LC-MS/MS. These analyses confirm that metabolite identification can be significantly improved by estimating the element composition in formulas using stable isotope labeling, or by introducing LC retention time during a spectral library search, which are incorporated into JUMPm functions. Finally, we compared the performance of JUMPm and two commonly used programs, Compound Discoverer 3.1 and MZmine 2, with respect to putative metabolite identifications. Our results indicate that JUMPm is an effective tool for metabolite identification of both unlabeled and labeled data in untargeted metabolomics.

Retention time shift analysis and correction in chemical isotope labeling liquid chromatography/mass spectrometry for metabolome analysis.

In chemical isotope labeling (CIL) liquid chromatography/mass spectrometry (LC/MS) metabolome analysis, the peak pairs of the same metabolite detected from different samples are aligned according to their mass and retention time (RT). Any RT shift of a peak pair in one of the sample files that falls outside the tolerance window will result in misalignment of the pair as a different metabolite. Thus, determination and correction of any significant RT shift are important to ensure the generation of high-quality metabolome results. In CIL LC/MS, the heavy-isotope-labeled pooled sample is spiked into all light-isotope-labeled individual samples. As a result, in the analysis of labeled samples of the same type, many common metabolites are detectable with high intensity in all LC/MS runs. We have developed a method to select a few of these metabolites as internal RT reference markers to check the occurrence of any RT shift in an LC/MS run. If a significant shift is found, an expanded list of these markers with their RT values covering the entire LC RT window is selected to serve as internal RT calibrants to recalibrate the chromatogram to correct any RT shift. We developed a software program in R to perform RT check (RTC) and recalibration (RT-calib). This program can quickly determine the occurrence of any RT shift falling outside a user-defined threshold in an LC/MS run, thereby triggering a timely intervention to correct the problem (e.g., fixing a small leak or changing a column). In the analysis of 278 dansylation LC/MS runs of human urine samples, we show that the RT values can be corrected to be within a 30-second window. An RT-check method and program tailored to CIL LC/MS metabolome analysis have been developed for quick detection and correction of RT shifts during the course of running many metabolome samples.

Abstract 4545: Quantitative profiling of Cytochrome P450 2S1 in colorectal cancer by PRM assay

Abstract Introduction: Cytochromes P450s (CYPs) constitute a superfamily of xenobiotic metabolising enzymes responsible for metabolism of many pharmaceuticals in the liver. Elevated mRNA levels of specific isoforms, such as CYP2S1, are associated with poor prognosis in colorectal cancer (CRC), and represent novel therapeutic targets for biotransformation of prodrugs to potent cytotoxics at the cancer site. In order to understand the expression of CYP2S1 protein, we have developed a parallel reaction monitoring mass spectrometry (PRM MS) assay to screen CRC samples. Method: Peptides (n=3) uniquely associated with CYP2S1 were synthesized and used as standards to optimise analytical performance (fragmentation conditions, LC retention time, LOQ, LOD, dynamic range) in an Orbitrap Fusion-based PRM MS assay. Levels of CYP2S1 were then determined in protein extracts of CRC, relative to the standards. Results: The PRM MS assay yielded quantitative data over 3 orders of magnitude. CYP2S1 was detected in C106, CaCO2, HCC2998, HT55 and DLD1 cell lines (0.05-1.08pg) and HT55 and DLD1 xenografts (0.29-0.41pg). CYP2S1 was also detected in patient-specific CRC tissues, with elevated levels in Stage III tumours. Conclusions: The PRM MS assay provides a specific, sensitive, high throughput method for identifying CYP2S1 compared to established enzyme and immunoassays. CYP2S1 levels vary considerably across CRC sources highlighting the importance of screening to identify the correct models for new drug development and the right patients for subsequent treatment. Citation Format: Sadr ul Shaheed, Laurence H. Patterson, Klaus Pors, Chris W. Sutton. Quantitative profiling of Cytochrome P450 2S1 in colorectal cancer by PRM assay [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 4545.

Centrifugal partition chromatography as a fractionation tool for the analysis of lignocellulosic biomass products by liquid chromatography coupled to mass spectrometry

The conversion of lignocellulosic biomass into biofuels and bio-products leads to oxygenated matrices having a wide range of polarities and molecular weights. A complete analytical characterization of these complex mixtures is necessary to improve conversion processes. In this study, an innovative centrifugal partition chromatography (CPC) protocol was developed to fractionate aqueous biomass samples with a MTBE-water solvent system, by mixing elution, displacement and extrusion modes in the same run. This new protocol was validated on model molecules and applied to the water soluble phase of a fast pyrolysis bio-oil. It demonstrated a promising separation with a relevant selectivity on the most significant chemical families of biomass samples: carbohydrates, furans, carboxylic acids and phenols. CPC fractions of the sample were collected and analyzed comprehensively by HPLC-UV/MS (with ESI negative and positive ionization modes). This CPC x LC approach allowed more accurate attributions on the 217 peaks detected. The use of different detection modes gave a complete view of the water soluble phase of a fast pyrolysis bio-oil through 2D maps. Molecular characterization was enhanced by independent information: CPC retention time, LC retention time, UV and MS spectra. Concomitance of these different chemical information is of precious help for unambiguous identification.

Comprehensive and Empirical Evaluation of Machine Learning Algorithms for Small Molecule LC Retention Time Prediction.

Liquid chromatography is a core component of almost all mass spectrometric analyses of (bio)molecules. Because of the high-throughput nature of mass spectrometric analyses, the interpretation of these chromatographic data increasingly relies on informatics solutions that attempt to predict an analyte's retention time. The key components of such predictive algorithms are the features these are supplies with, and the actual machine learning algorithm used to fit the model parameters. Therefore, we have evaluated the performance of seven machine learning algorithms on 36 distinct metabolomics data sets, using two distinct feature sets. Interestingly, the results show that no single learning algorithm performs optimally for all data sets, with different types of algorithms achieving top performance for different types of analytes or different protocols. Our results thus show that an evaluation of machine learning algorithms for retention time prediction is needed to find a suitable algorithm for specific analytes or protocols. Importantly, however, our results also show that blending different types of models together decreases the error on outliers, indicating that the combination of several approaches holds substantial promise for the development of more generic, high-performing algorithms.

Non-target data acquisition for target analysis (nDATA) of 845 pesticide residues in fruits and vegetables using UHPLC/ESI Q-Orbitrap.

A non-target data acquisition for target analysis (nDATA) workflow based on accurate mass measurements using UHPLC/ESI Q-Orbitrap full MS-data-independent acquisition and a compound database was developed to screen pesticide residues in fruit and vegetable samples. The compound database of 845 pesticides was built from dd-MS2 (data-dependent acquisition) product ion spectral data and LC retention times of individual pesticide standards. MS2 spectra of samples were acquired using multiplexing data-independent acquisition (mDIA) and variable data-independent acquisition (vDIA). Screening of pesticides in samples was based on either the retention time (± 0.5min) and the mass accuracy (± 5ppm) of a precursor (RTP by full MS) or the retention time (± 0.5min) and the mass accuracy (± 5ppm) of a precursor and its fragment ion (RTFI by full MS/DIA). In validation studies involving mDIA and vDIA analysis of 10 fruits and vegetables spiked with pesticides prior to QuEChERS sample preparation, RTP correctly found up to 765 and 796 pesticides at 10 and 100μg/kg, respectively, whereas RTFI correctly identified up to 729 and 764 pesticides at the same respective concentrations. UHPLC/ESI Q-Orbitrap full MS/mDIA or vDIA proved to be a comprehensive detection technique and has potential for pesticide residue screening in fruits and vegetables. Graphical Abstract.