Method for metabolomic analysis of biological samples by integrating dynamic parameters of nmr

Abstract

The invention relates to a method for metabolomic analysis of a biological sample, which comprises the following steps: - carrying out nuclear magnetic resonance (NMR) spectroscopy of the sample in order to determine a concentration spectrum (2) of the metabolites included in the sample and a measurement of at least one dynamic parameter of the metabolites; - discretising the spectra in order to obtain a concentration data matrix and at least one dynamic parameter data matrix; and - establishing a comparison, for example by principal component analysis, between the data of the matrices and the reference data.

Description

Method for metabolomic analysis of biological samples by integration of dynamic NMR parameters

[0001] The invention relates to metabolomics. More particularly, the invention relates to a method for metabolomic analysis of biological samples which makes it possible to increase the discriminatory power of the method by integrating dynamic NMR parameters into the statistical processing, as well as technical means allowing the implementation of such a method.

[0002] Metabolomics aims to characterize and identify biomarkers produced, used, and excreted by cells, organs, or organisms, associated with defined systemic states. Common applications of metabolomics include identifying human health conditions, diagnosing diseases, and testing drug efficacy and toxicity. As a more specific example, metabolomics can be applied in cancer diagnosis by analyzing tissues or physiological fluid and comparing them with healthy reference samples or a reference database.

[0003] It is known from the state of the art to carry out such an analysis by performing nuclear magnetic resonance spectroscopy, commonly referred to by the acronym NMR, of the sample to determine a concentration spectrum of the metabolites included in the sample. This measurement technique makes it possible to identify the metabolites included in the sample and to determine their concentration, in a manner known per se. The spectrum obtained is then processed and compared to reference data, generally derived from a similar spectrum obtained by NMR spectroscopy of a healthy sample or from a pre-established reference database, with the aim of identifying any significant differences between the two spectra and obtaining a diagnosis. Document WO 2011 041892 A1 describes the implementation of such an analysis.

[0004] This type of analysis allows interesting results to be obtained but may present limitations in the exploitation that can be made of the results of the analysis.

[0005] The invention aims in particular to improve the robustness of the metabolomic analysis method.

[0006] To this end, the invention relates to a method for metabolomic analysis of a biological sample, in which the following steps are implemented:

[0007] - nuclear magnetic resonance (NMR) spectroscopy of the sample is carried out to determine a concentration spectrum of the metabolites included in the sample and a spectrum of at least one dynamic parameter of the metabolites,

[0008] - the spectra are discretized to obtain a concentration data matrix and at least one dynamic parameter data matrix, and

[0009] - a comparison is carried out, by a multivariate statistical analysis, for example by means of a principal component analysis (PCA) or supervised (orthogonal projection discriminant analysis, OPLS-DA), advantageously between the data of the matrices and the reference data.

[0010] Thus, in addition to obtaining and exploiting an NMR spectrum in concentration of the sample metabolites in accordance with the conventional approach described above, at least one additional spectrum is obtained and exploited to characterize one or more dynamic parameters of the metabolites providing information on their molecular movements and the interactions they undergo within the sample. This additional information makes it possible to refine the metabolomic analysis carried out by increasing its discriminatory power and therefore its robustness. Thus, by combining quantitative (concentration) and dynamic data, it is possible not only to increase the discriminatory power of certain metabolites but also to reveal significant metabolites by their dynamics whereas they were not significant by their concentrations considered in isolation.Indeed, we understand that the multiplication of measured data, and implicitly of reference data, allows for a more detailed comparison between the data of a sample and reference data.

[0011] Advantageously, nuclear magnetic resonance (NMR) spectroscopy of the sample is carried out to determine a spectrum of at least one of the dynamic parameters of the metabolites from the following list: diffusion coefficient (D), first longitudinal relaxation time (T1), second transverse relaxation time (T2), third longitudinal relaxation time in the rotating frame (T 1rho ).

[0012] The at least one dynamic parameter can thus be easily measured by nuclear magnetic resonance spectroscopy, as can the concentration data, so that the method does not require considerable additional technical means for its implementation. In other words, this method does not require significantly more technical resources than the conventional method of the prior art. Furthermore, the parameters in the list correspond to parameters that make it possible to effectively characterize the dynamic aspects of the metabolites included in the sample.

[0013] Preferably, nuclear magnetic resonance (NMR) spectroscopy of the sample is performed to measure several dynamic parameters from the list, or even all of the parameters from the list.

[0014] This further improves the robustness of the metabolomic analysis process.

[0015] According to a first embodiment of the invention, during the step of carrying out the comparison, the concentration data and the dynamic parameter data are compared, as such, with the reference data.

[0016] According to a second embodiment of the invention, after the discretization step, the concentration data matrix and the at least one dynamic parameter data matrix are assembled to obtain a global data matrix. In this case, several statistical treatments can be envisaged, such as for example multi-block processing.

[0017] According to a third embodiment of the invention, after the discretization step, the concentration data matrix is ​​weighted by the at least one dynamic parameter data matrix. This weighting can also be obtained directly by NMR spectroscopy by applying relaxation or diffusion filters to obtain a concentration spectrum weighted by a dynamic parameter.

[0018] In the first embodiment, the data from each of the spectra are compared with as much reference data. In the second embodiment, the data are first grouped, then these grouped data are compared with reference data. In the third embodiment, the concentration data are weighted by the dynamic parameter data before comparison with reference data. The data from the NMR spectra can thus be used in several different ways, which provides a certain flexibility in implementing the invention.

[0019] Also provided according to the invention is a computer program comprising instructions for causing an NMR spectrometer comprising a processor to execute the steps of a method as defined above.

[0020] The invention also provides a computer-readable recording medium on which the computer program as defined above is recorded. Brief description of the figures

[0021] The invention will be better understood on reading the following description, given solely by way of example and with reference to the appended drawings in which:

[0022] is a graphical representation of an NMR spectrum obtained during the implementation of a metabolomic analysis method according to the invention,

[0023] illustrates data matrices obtained by discretization of spectra such as that of the, and

[0024] is a graphical representation of a comparison step by principal component analysis based on the data from the matrices of the. Detailed description

[0025] The result of a first step of a method for metabolomic analysis of a biological sample according to the invention is shown. This biological sample is for example a sample of physiological fluid, in this case a sample containing cancer cell lines originating from breast cancer. In this first step, nuclear magnetic resonance (NMR) spectroscopy of the sample is carried out to determine a concentration spectrum 2 of the metabolites included in the sample and a spectrum with a measurement of at least one dynamic parameter of the metabolites. The dynamic parameter is here chosen from the following list: diffusion coefficient D, first longitudinal relaxation time T1, second transverse relaxation time T2, third longitudinal relaxation time in the rotating frame T 1rho These spectra obtained by spectroscopy of the sample will be referred to as “sample spectra”.

[0026] Alternatively, such spectroscopies are also performed with reference biological samples, or one or more reference spectra are imported from a pre-existing database. In both cases, in the following, any spectrum actually obtained by spectroscopy of a reference sample or imported from the database will be referred to as a "reference spectrum".

[0027] Illustrates the concentration spectrum 2 obtained by implementing such spectroscopy, the graphical representation of which presents the chemical shift δ on the abscissa and the amplitude A of the signal on the ordinate. In a manner known per se, the presence of a peak at a certain chemical shift value is attributed to the presence of a metabolite determined in concentration proportional to the amplitude of the signal. In the same way, for a spectrum in dynamic parameter chosen from the aforementioned list, the presence of a peak at a certain chemical shift value is attributed to the presence of a determined metabolite for which the value of the dynamic parameter is measured with the techniques available in the library of pulse sequences of NMR spectrometers.

[0028] After implementing this first step, we have a sample spectrum in concentration, one or more sample spectra in dynamic parameter. We also have preference spectra in concentration and mxpreference spectra in dynamic parameter, where p is a non-zero natural integer and m corresponds to the number of dynamic parameters from the list retained, m being between 1 and 4 in this case. We then proceed to a second step of the metabolomic analysis process during which we discretize each of the spectra. As illustrated in the, the discretization consists of separating the abscissa axis into a finite number of intervals 4 allowing the entire spectrum to be covered. We assign to each of these intervals 4 a central chemical shift value and a signal value obtained by integrating the curve over the interval.The same applies to dynamic parameters, where each interval is assigned at least one value of a dynamic parameter. This discretization facilitates the processing of spectra data.

[0029] The result of a second step of the metabolomic analysis process is represented. In this second step, a concentration matrix 6a is constructed in which each row corresponds to one of the spectra denoted S1, S2, …etc., the sample spectrum in concentration or one of the reference spectra in concentration, and each column corresponds to one of the intervals 4 constructed during the discretization of the spectra. Each row / column intersection is filled with the value of the integral of the signal intensity of the spectrum over the interval in question, transformed in a manner known per se into the concentration value. By construction, the concentration matrix 6a comprises n columns and p + 1 rows.

[0030] In the same way, we construct dynamic parameter matrices 6. For each dynamic parameter retained, the dynamic parameter matrix is ​​such that each row corresponds to one of the spectra, the sample spectrum in dynamic parameter or one of the reference spectra in dynamic parameter, and each column corresponds to one of the intervals 4 constructed during the discretization of the spectra. Each row / column intersection is filled with the value of the dynamic parameter. By construction, we obtain dynamic parameter matrices 6, each of them comprising n columns and p + 1 rows. In the embodiment as illustrated in, two dynamic parameter matrices 6 were constructed: a matrix 6b of first longitudinal relaxation time T1 and a matrix 6c of diffusion coefficient D.

[0031] The result of a third step of the metabolomic analysis method is shown. In this third step, a comparison is carried out by principal component analysis between the data of the matrices 6 corresponding to the sample spectra and the data of the matrices 6 corresponding to the reference spectra. This is a graphical representation of an operation commonly referred to by the English expression "score plotting" and widely known in the field of statistical analysis, so that it will not be described further in the following. Using this comparison, it is possible to qualify a state of the sample, in particular in one of the applications mentioned in the preamble of the present patent application. From a technical point of view, the comparison within the framework of the invention can be implemented in several different ways.

[0032] According to a first embodiment of the invention, during the step of performing the comparison, the concentration data and the dynamic parameter data are compared, as such, with the reference data. This is a simple method to implement and does not require additional processing of the matrices.

[0033] According to a second embodiment of the invention, after the discretization step, the concentration data matrix and the at least one dynamic parameter data matrix are assembled to obtain a global data matrix. According to this method, a single comparison operation is carried out, for example a single operation called "score plotting", which reduces the processing time linked to the implementation of the comparison operation.

[0034] According to a third embodiment of the invention, after the discretization step, the concentration data matrix is ​​weighted by the at least one dynamic parameter data matrix. This weighting makes it possible to refine the data obtained. From a mathematical point of view, to carry out this weighting, the values ​​of each cell of the concentration data matrix are multiplied or divided by the value(s) of the dynamic parameter data matrix(es) contained in the same cell. In other words, for any pair (i; j) defining an existing cell of the concentration matrix, the concentration value of the cell (i; j) is multiplied or divided by the value(s) of the cell (i; j) of the dynamic parameter data matrix(es).

[0035] The invention is not limited to the embodiments presented and other embodiments will become apparent to those skilled in the art.

[0036] It is expected that several samples will be subjected to NMR spectroscopy. If q is the number of samples tested, then the matrices include p + q rows.

[0037] The invention is applicable to the analysis of samples other than biological samples. Examples include food samples, for example honey or animal products leaving the slaughterhouse, and samples relating to materials science for carrying out quality control. List of references

[0038] 2: concentration spectrum4: interval6a, 6b, 6c: matrix8: principal component analysis

Claims

Method for metabolomic analysis of a biological sample, characterized in that the following steps are implemented:- nuclear magnetic resonance (NMR) spectroscopy of the sample is carried out to determine a concentration spectrum (2) of the metabolites included in the sample and a spectrum with a measurement of at least one dynamic parameter of the metabolites,- the spectra are discretized to obtain a matrix (6a) of concentration data and at least one matrix (6b, 6c) of dynamic parameter data, and- a comparison is carried out, by a multivariate statistical analysis, between the data of the matrices and the reference data. Method for metabolomic analysis of a biological sample according to the preceding claim, in which the comparison is carried out by means of a principal component analysis (8) or supervised. Method for metabolomic analysis of a biological sample according to any one of the preceding claims, in which nuclear magnetic resonance (NMR) spectroscopy of the sample is carried out to determine a spectrum of at least one of the dynamic parameters of the metabolites from the following list: diffusion coefficient (D), first longitudinal relaxation time (T1), second transverse relaxation time (T2), third longitudinal relaxation time in the rotating frame (T 1rho ). Method for metabolomic analysis of a biological sample according to the preceding claim, in which nuclear magnetic resonance (NMR) spectroscopy of the sample is carried out to measure several dynamic parameters from the list, or even all the parameters from the list. A method for metabolomic analysis of a biological sample according to any one of claims 2 to 4, wherein during the step of performing the comparison, the concentration data and the dynamic parameter data are compared, as such, to the reference data. Method for metabolomic analysis of a biological sample according to any one of claims 1 to 4, in which after the discretization step, the concentration data matrix (6a) and the at least one dynamic parameter data matrix (6b, 6c) are assembled to obtain a global data matrix. Method for metabolomic analysis of a biological sample according to any one of claims 1 to 4, in which after the discretization step, the concentration data matrix (6a) is weighted by the at least one dynamic parameter data matrix (6b, 6c). A computer program comprising instructions for causing an NMR spectrometer comprising a processor to perform the steps of a method according to any one of claims 1 to 7. Computer-readable recording medium on which the computer program according to claim 8 is recorded.

Images