On mass spectrometer instrument standardization and interlaboratory calibration transfer using neural networks

Standardization method, testing scenario, and accuracy of the infrared prediction model affect the standardization accuracy of milk mid-infrared spectra

The widespread use of milk mid-infrared (MIR) spectroscopy for phenotype prediction has urged the application of prediction models across regions and countries. Spectra standardization is the most effective way to reduce the variability in the spectral signal provided by different instruments and labs. This study aimed to develop different standardization models for MIR spectra collected by multiple instruments, across 2 provinces of China, and investigate whether the standardization method (piecewise direct standardization, PDS, and direct standardization, DS), testing scenario (standardization of spectra collected on the same day or after 7 mo), infrared prediction model accuracy (high or low), and instrument (6 instruments from 2 brands) affect the performance of the standardization model. The results showed that the determination coefficient (R2) between absorbance values at each wavenumber provided by the primary and the secondary instruments increased from less than 0.90 to nearly 1.00 after standardization. Both PDS and DS successfully reduced spectra variation among instruments, and performed significantly better than non-standardization (P < 0.05). However, DS was more prone to overfitting than PDS. Standardization accuracy was higher when tested using spectra collected on the same time compared with those collected 7 mo after (P < 0.05), but great improvement in model transferability was obtained for both scenarios compared with the non-standardized spectra. The less accurate infrared prediction model (for C8:0 and C10:0 content) benefited the most (P < 0.05) from spectra standardization compared with the more accurate model (for total fat and protein content). For spectra collected after 7 mo from standardization, after PDS the RMSE between predictions obtained by different machines decreased on average by 86 and 94% compared with the values before standardization, for C8:0 and C10:0 respectively. The secondary instrument had no significant effect on the R2 between predictions (P > 0.05). The variation in the spectral signal provided by different instruments was successfully reduced by standardization across 2 provinces in China. This study lays the foundations for developing a national MIR spectra database to provide consistent predictions across provinces to be used in dairy farm management and breeding programs in China. Besides, this provides opportunities for data exchange and cooperation at international levels.

The application of artificial neural networks in metabolomics: a historical perspective

BACKGROUND

Metabolomics data, with its complex covariance structure, is typically modelled by projection-based machine learning (ML) methods such as partial least squares (PLS) regression, which project data into a latent structure. Biological data are often non-linear, so it is reasonable to hypothesize that metabolomics data may also have a non-linear latent structure, which in turn would be best modelled using non-linear equations. A non-linear ML method with a similar projection equation structure to PLS is artificial neural networks (ANNs). While ANNs were first applied to metabolic profiling data in the 1990s, the lack of community acceptance combined with limitations in computational capacity and the lack of volume of data for robust non-linear model optimisation inhibited their widespread use. Due to recent advances in computational power, modelling improvements, community acceptance, and the more demanding needs for data science, ANNs have made a recent resurgence in interest across research communities, including a small yet growing usage in metabolomics. As metabolomics experiments become more complex and start to be integrated with other omics data, there is potential for ANNs to become a viable alternative to linear projection methods.

AIM OF REVIEW

We aim to first describe ANNs and their structural equivalence to linear projection-based methods, including PLS regression. We then review the historical, current, and future uses of ANNs in the field of metabolomics.

KEY SCIENTIFIC CONCEPT OF REVIEW

Is metabolomics ready for the return of artificial neural networks?

Standardization of complex biologically derived spectrochemical datasets

Spectroscopic techniques such as Fourier-transform infrared (FTIR) spectroscopy are used to study interactions of light with biological materials. This interaction forms the basis of many analytical assays used in disease screening/diagnosis, microbiological studies, and forensic/environmental investigations. Advantages of spectrochemical analysis are its low cost, minimal sample preparation, non-destructive nature and substantially accurate results. However, an urgent need exists for repetition and validation of these methods in large-scale studies and across different research groups, which would bring the method closer to clinical and/or industrial implementation. For this to succeed, it is important to understand and reduce the effect of random spectral alterations caused by inter-individual, inter-instrument and/or inter-laboratory variations, such as variations in air humidity and CO₂ levels, and aging of instrument parts. Thus, it is evident that spectral standardization is critical to the widespread adoption of these spectrochemical technologies. By using calibration transfer procedures, in which the spectral response of a secondary instrument is standardized to resemble the spectral response of a primary instrument, different sources of variation can be normalized into a single model using computational-based methods, such as direct standardization (DS) and piecewise direct standardization (PDS); therefore, measurements performed under different conditions can generate the same result, eliminating the need for a full recalibration. Here, we have constructed a protocol for model standardization using different transfer technologies described for FTIR spectrochemical applications. This is a critical step toward the construction of a practical spectrochemical analysis model for daily routine analysis, where uncertain and random variations are present.

Transfer of multivariate regression models between high-resolution NMR instruments: application to authenticity control of sunflower lecithin

In recent years the number of spectroscopic studies utilizing multivariate techniques and involving different laboratories has been dramatically increased. In this paper the protocol for calibration transfer of partial least square regression model between high-resolution nuclear magnetic resonance (NMR) spectrometers of different frequencies and equipped with different probes was established. As the test system previously published quantitative model to predict the concentration of blended soy species in sunflower lecithin was used. For multivariate modelling piecewise direct standardization (PDS), direct standardization, and hybrid calibration were employed. PDS showed the best performance for estimating lecithin falsification regarding its vegetable origin resulting in a significant decrease in root mean square error of prediction from 5.0 to 7.3% without standardization to 2.9-3.2% for PDS. Acceptable calibration transfer model was obtained by direct standardization, but this standardization approach introduces unfavourable noise to the spectral data. Hybrid calibration is least recommended for high-resolution NMR data. The sensitivity of instrument transfer methods with respect to the type of spectrometer, the number of samples and the subset selection was also discussed. The study showed the necessity of applying a proper standardization procedure in cases when multivariate model has to be applied to the spectra recorded on a secondary NMR spectrometer even with the same magnetic field strength. Copyright © 2016 John Wiley & Sons, Ltd.

Maintaining the predictive abilities of multivariate calibration models by spectral space transformation

In quantitative on-line/in-line monitoring of chemical and bio-chemical processes using spectroscopic instruments, multivariate calibration models are indispensable for the extraction of chemical information from complex spectroscopic measurements. The development of reliable multivariate calibration models is generally time-consuming and costly. Therefore, once a reliable multivariate calibration model is established, it is expected to be used for an extended period. However, any change in the instrumental response or variations in the measurement conditions can render a multivariate calibration model invalid. In this contribution, a new method, spectral space transformation (SST), has been developed to maintain the predictive abilities of multivariate calibration models when the spectrometer or measurement conditions are altered. SST tries to eliminate the spectral differences induced by the changes in instruments or measurement conditions through the transformation between two spectral spaces spanned by the corresponding spectra of a subset of standardization samples measured on two instruments or under two sets of experimental conditions. The performance of the method has been tested on two data sets comprising NIR and MIR spectra. The experimental results show that SST can achieve satisfactory analyte predictions from spectroscopic measurements subject to spectrometer/probe alteration, when only a few standardization samples are used. Compared with the existing popular methods designed for the same purpose, i.e. global PLS, univariate slope and bias correction (SBC) and piecewise direct standardization (PDS), SST has the advantages of implementation simplicity, wider applicability and better performance in terms of predictive accuracy.

Mass spectrometry based sensor strategies for the authentication of oysters according to geographical origin

This study was undertaken to investigate the relevance of using the pyrolysis-MS (Py-MS) technique to discriminate the production area of oysters harvested over two years and to assess from the data of the second year of harvest the potential of an alternative MS-based technique, the solid phase microextraction-MS (SPME-MS), to perform this discrimination. Oysters were harvested in various areas of France, and models of discrimination according to harvest season were built from Py-MS fingerprints and from virtual SPME-MS fingerprints obtained by summing the mass spectra generated by the SPME-GC-MS system. The treatment of the Py-MS data by a 21-12-3 artificial neural networks led to a correct classification of only 89.2% of the oyster samples according to shoreline. The misclassifications thus did not allow use of the Py-MS technique as a relevant tool for authentication of oyster origin. The assessment of the potential of the virtual SPME-MS fingerprints to discriminate the production area of oysters was undertaken on a part of the sample set. The virtual SPME-MS data were pretreated according to two methods, filtering of raw data (FRD) and comprehensive combinatory standard correction (CCSC), a recently developed chemometric method used for the correction of instrumental signal drifts in MS systems. The results obtained with the virtual SPME-MS fingerprints are promising because this technique, when the data were pretreated by the CCSC method, led to a successful discrimination of the oyster samples not only according to shoreline but also according to production region. This study confirms that an efficient correction method (CCSC) of instrumental drifts can considerably increase the discriminative information contained in the volatile fraction of food products.

Chemometric study on the TiO2-photocatalytic degradation of nitrilotriacetic acid

A chemometric study on the TiO2-photocatalytic degradation of nitrilotriacetic acid (NTA) in aqueous media under UV radiation has been carried out taking into account the multiple variables that take part in the system. To save redundant number of experiments, the system has been managed under chemometric techniques for several variables as NTA and TiO2 concentrations, pH and irradiation time. Multiple-way analysis of the variance (MANOVA) has been applied to find the statistically significant variables. An artificial neural network (ANN) has been used to build an empirical model of the system. All measurements have been driven under experimental designs: a full-factorial design (FFD) was used to analyze significant factors through MANOVA, and a Doehlert design, which was modified by spatial rotation, was applied in order to have a satisfactory number of levels for the factor time to be able to train the ANN. The study allows the knowledge and prediction of the behavior of the system as well as to work out kinetic parameters and to optimize their variables. The results of kinetic parameters obtained with the neural network agreed with independent experimental results, confirming a Langmuir-Hinshelwood kinetic regime. The difference between NTA and ethylenediaminetetraacetic acid (EDTA), which has been previously studied, is also established.

Metabolomics: current technologies and future trends

The ability to sequence whole genomes has taught us that our knowledge with respect to gene function is rather limited with typically 30-40% of open reading frames having no known function. Thus, within the life sciences there is a need for determination of the biological function of these so-called orphan genes, some of which may be molecular targets for therapeutic intervention. The search for specific mRNA, proteins, or metabolites that can serve as diagnostic markers has also increased, as has the fact that these biomarkers may be useful in following and predicting disease progression or response to therapy. Functional analyses have become increasingly popular. They include investigations at the level of gene expression (transcriptomics), protein translation (proteomics) and more recently the metabolite network (metabolomics). This article provides an overview of metabolomics and discusses its complementary role with transcriptomics and proteomics, and within system biology. It highlights how metabolome analyses are conducted and how the highly complex data that are generated are analysed. Non-invasive footprinting analysis is also discussed as this has many applications to in vitro cell systems. Finally, for studying biotic or abiotic stresses on animals, plants or microbes, we believe that metabolomics could very easily be applied to large populations, because this approach tends to be of higher throughput and generally lower cost than transcriptomics and proteomics, whilst also providing indications of which area of metabolism may be affected by external perturbation.

Gene expression analysis and clinical diagnosis

BACKGROUND

A new basis for diagnostic tests is being provided by the vast amount of data on gene expression that are now becoming available through large-scale measurement of mRNA abundance. The insights gained from these resources are most likely going to provide both a better basic understanding of disease mechanisms, and to identify molecular markers for more precise diagnoses and for prediction of prognosis and treatment response.

METHODS

Some quantitative RT-PCR assays are utilized today for diagnosis of both malignant and non-malignant disease, but the use of gene expression measurements in clinical medicine can be expected to increase dramatically.

CONCLUSIONS

There are important technical issues that must be adequately solved in order to obtain robust assays, such as standardized protocols with appropriate quality controls that ensure reliable data for the specific samples being analysed and good inter-laboratory reproducibility.

Simultaneous resolution of overlapping peaks in high-performance liquid chromatography and micellar electrokinetic chromatography with diode array detection using augmented iterative target transformation factor analysis

In this paper, augmentation has been applied to data matrices, which originate from hyphenated methods that share the same mode of detection, but use different separation methods, HPLC-DAD and MEKC-DAD. A novel method, wavelength shift eigenstructure tracking (WET), has been proposed for the alignment between the wavelength scale of both detectors. WET proves to be suitable for the detection as well as correction of wavelength shift between both detectors. After correction of the wavelength scale, data obtained on both systems have been augmented and submitted to iterative target transformation factor analysis. Augmented curve resolution provides significantly better estimates of the chromatographic and electrophoretic profiles and spectra than the use of non-augmented curve resolution on HPLC and MEKC data separately. It is particularly useful when the pure fraction of a chromatographic peak is less than 0.10. Finally, the relative weight of MEKC versus HPLC in augmentation may be increased using intensity and noise normalisation. However, since noise normalisation and its accompanying decrease in signal-to-noise ratio leads to a loss of information, and, since intensity normalisation may cause a failure of the augmented curve resolution algorithm, benefits and drawbacks of normalisation should be weighed on a case-by-case basis.

Monitoring of enzymatic hydrolysis of penicillin G by pyrolysis-negative ion mass spectrometry

A pyrolysis-negative ion mass spectrometry (Pyr-NIMS) is used for the monitoring of enzymatic hydrolysis of penicillin G (Pen G) to 6-aminopenicillanic acid (6-APA) and phenyl acetic acid (PAA). The high sensitivity and rapid response time of Pyr-NIMS allow its application to the simultaneously determination of these compounds. The mass to charge (m/z) values of 262, 156 and 135 of Pen G, 6-APA and PAA respectively, are used for the quantitative measurements by selected ion monitoring (SIM). The limit of detection (LOD), linearity and relative standard deviation (n=5) are 10 ng ml(-1), 100 ng ml(-1)-1000 mg ml(-1) and 1.5%, respectively The results are compared with high performance liquid chromatography (HPLC). An important advantage of the presented analytical system is the high linearity of signals without preliminary separation and recalibration. The main and interactive effects of pH, temperature and concentration of Pen G for enzymatic hydrolysis of Pen G are studied. Optimize conditions of pH (8), temperature (28 degrees C) and concentration of Pen G (12% w/v) in real samples are obtained.

Optimization and empirical modeling of HG-ICP-AES analytical technique through artificial neural networks

An artificial neural network technique has been applied to the optimization of a hydride generation-inductively coupled plasma-atomic emission spectrometry (HG-ICP-AES) coupling for the determination of Ge at trace levels. The back propagation of errors net architecture was used. Experimental parameters and their relationship have been studied, obtaining a surface response of the system. The results and optimization aspects achieved with the neural network approach have been compared to the "one variable at time" and SIMPLEX methods.

A functional genomics strategy that uses metabolome data to reveal the phenotype of silent mutations

A large proportion of the 6,000 genes present in the genome of Saccharomyces cerevisiae, and of those sequenced in other organisms, encode proteins of unknown function. Many of these genes are "silent, " that is, they show no overt phenotype, in terms of growth rate or other fluxes, when they are deleted from the genome. We demonstrate how the intracellular concentrations of metabolites can reveal phenotypes for proteins active in metabolic regulation. Quantification of the change of several metabolite concentrations relative to the concentration change of one selected metabolite can reveal the site of action, in the metabolic network, of a silent gene. In the same way, comprehensive analyses of metabolite concentrations in mutants, providing "metabolic snapshots," can reveal functions when snapshots from strains deleted for unstudied genes are compared to those deleted for known genes. This approach to functional analysis, using comparative metabolomics, we call FANCY-an abbreviation for functional analysis by co-responses in yeast.