Method for predicting retention time

Abstract

Predict the retention time of the target substance obtained by high-performance liquid chromatography. [Solution] In the retention time prediction method, a parametric function is created using a linear combination of the partition coefficients between the mobile phase and multiple stationary phase models. For multiple standard substances, the partition coefficients between the mobile phase and multiple stationary phase models are calculated by quantum chemical calculations using molecular structure information. Using the retention times and partition coefficients of the multiple standard substances, the parameters of the partition coefficients in the linear combination of the parametric function are determined by linear regression. A specific function is created by substituting the parameters into the parametric function. For the target substance, the partition coefficients between the mobile phase and multiple stationary phase models are calculated by quantum chemical calculations using molecular structure information. Using the specific function, the retention time is calculated from the partition coefficients for the target substance.

Description

[Technical Field]

[0001] This invention relates to a method for predicting the retention time of a target substance obtained by high-performance liquid chromatography. [Background technology]

[0002] Retention times obtained by high-performance liquid chromatography (HPLC) are widely used for identifying the components of target substances. Retention times in HPLC are thought to depend on the partition coefficient between the mobile phase and the stationary phase. Patent documents 1-3 disclose techniques for evaluating the octanol-water partition coefficient based on retention times. [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2001-124756 [Patent Document 2] Japanese Patent Publication No. 2001-228130 [Patent Document 3] Japanese Patent Publication No. 2002-267647 [Non-patent literature]

[0004] [Non-Patent Document 1] E. Choi, W. J. Yoo, H.-Y. Jang, T.-Y. Jamg, SK Lee, HB Oh: J. Chromatogr. A, 1705, 464167 (2023). [Overview of the Initiative] [Problems that the invention aims to solve]

[0005] However, the relationship between retention time and the octanol-water partition coefficient is treated logarithmically (logP) (or as a log-log plot), making it difficult to treat as a real number.

[0006] If retention time could be predicted from the partition coefficient, it would be possible to easily identify the components of a sample even when standard materials are not available. However, predicting retention time from the partition coefficient is not easy because the mechanism by which molecules constituting the target substance are retained in the mobile and stationary phases involves multiple interactions, including not only hydrophobic interactions with octadecyl groups, but also electrostatic interactions and hydrogen bonding interactions with silanol groups.

[0007] Non-patent document 1 discloses a technique for predicting retention time using machine learning. In this technique, dansylation is performed as a derivatization treatment of the target substance, so that the mechanism by which the molecules constituting the target substance are retained in the mobile phase and stationary phase can be handled solely by hydrophobic interactions. Although this technique has high accuracy in predicting retention time, it is not practical to perform derivatization treatment on unknown components.

[0008] Currently available prediction software and systems include ACD / ChromGenius, DryLab, and ChromSword. However, these software programs generally make predictions based on actual measurement data using standard materials or specific analytical conditions, and therefore still have challenges in terms of versatility in accurately predicting retention times across a wide range of compounds and measurement conditions.

[0009] The objective of this invention is to predict the retention time of a target substance obtained by high-performance liquid chromatography. [Means for solving the problem]

[0010] In one embodiment of the present invention, the prediction method predicts the retention time of a target substance obtained by high-performance liquid chromatography. The solvent composition of the mobile phase is determined. Multiple stationary phase models are defined. A parametric function is constructed using a linear combination of the distribution coefficients between the stationary phase and the mobile phase in the aforementioned multiple stationary phase models. Retention times and molecular structure information of a plurality of standard substances are obtained. For the plurality of standard substances, partition coefficients between the mobile phase in the plurality of stationary phase models are calculated by quantum chemical calculations using the molecular structure information. Using the retention times and partition coefficients of the plurality of standard substances, parameters of the partition coefficients in the primary bond of the parametric function are determined by linear regression. A specific function is created by substituting the parameters into the parametric function. Molecular structure information of the target substance is obtained. For the target substance, partition coefficients between the mobile phase in the plurality of stationary phase models are calculated by quantum chemical calculations using the molecular structure information. Using the specific function, the retention time is calculated from the partition coefficient for the target substance.

[0011] In the prediction method according to one embodiment of the present invention, the bioaccumulation of a target substance is predicted. A plurality of stationary phase models are defined. A parametric function using the primary bond of the partition coefficient between water in the plurality of stationary phase models is created. Bioaccumulation and molecular structure information of a plurality of standard substances are obtained. For the plurality of standard substances, partition coefficients between water in the plurality of stationary phase models are calculated by quantum chemical calculations using the molecular structure information. Using the bioaccumulation and partition coefficients of the plurality of standard substances, parameters of the partition coefficients in the primary bond of the parametric function are determined by linear regression. A specific function is created by substituting the parameters into the parametric function. Molecular structure information of the target substance is obtained. For the target substance, partition coefficients between water in the plurality of stationary phase models are calculated by quantum chemical calculations using the molecular structure information. Using the specific function, the bioaccumulation is calculated from the partition coefficient for the target substance. [Effects of the Invention]

[0012] According to the present invention, the retention time obtained by high-performance liquid chromatography for a target substance can be predicted. [Brief explanation of the drawing]

[0013] [Figure 1] This flowchart shows a method for predicting retention time according to one embodiment of the present invention. [Figure 2] This is a schematic model illustrating the mobile and stationary phases in a column during HPLC using isocratic elution. [Figure 3] This is a schematic model illustrating the mobile phase and stationary phase in an HPLC column using gradient elution. [Figure 4] This is a schematic model that shows only the mobile phase in the column during HPLC using gradient elution. [Figure 5] This graph shows the results obtained in Example 1 when the solvent composition (volume ratio) of the mobile phase was set to acetonitrile:water = 100:0. [Figure 6] This graph shows the results obtained in Example 1 when the solvent composition (volume ratio) of the mobile phase was set to acetonitrile:water = 90:10. [Figure 7] This graph shows the results obtained in Example 1 when the solvent composition (volume ratio) of the mobile phase was set to acetonitrile:water = 80:20. [Figure 8] This graph shows the results obtained in Example 1 when the solvent composition (volume ratio) of the mobile phase was acetonitrile:water = 60:40. [Figure 9] This graph shows the time change of the volume fraction φA of acetonitrile in the mobile phase in Example 2. [Figure 10] This graph shows the calculation results of logp(φA(t)) for each standard substance in Example 2. [Figure 11] This graph shows the results of fitting the quantum chemical calculations in Example 2. [Modes for carrying out the invention]

[0014] [Overall explanation] (Introduction) A retention time prediction method according to one embodiment of the present invention is configured to accurately predict the retention time obtained by high-performance liquid chromatography (HPLC) from the molecular structure of a substance of any molecular structure. In other words, with the retention time prediction method according to this embodiment, the retention time obtained by HPLC can be easily determined without actually performing HPLC.

[0015] In the retention time prediction method according to this embodiment, for example, the retention time of a substance can be predicted from the molecular structure of the substance expected to be contained in a sample to be analyzed by HPLC. Therefore, by using the retention time prediction method according to this embodiment, it becomes possible to identify components other than standard substances contained in any sample from the chromatogram obtained by HPLC. In other words, by using the retention time prediction method according to this embodiment, the amount of substances that can be identified by HPLC increases dramatically.

[0016] Figure 1 is a flowchart of the retention time prediction method according to this embodiment. The retention time prediction method according to this embodiment includes steps S01 to S10. Steps S01 to S10 will be described below with reference to Figure 1.

[0017] (Step S01: Determine the solvent composition of the mobile phase) Step S01 determines the solvent composition of the HPLC mobile phase. The solvent composition of the mobile phase determined in Step S01 can be arbitrarily determined depending on the sample to be analyzed by HPLC, for example. The solvent of the mobile phase is not particularly limited as long as it is a solvent commonly used as a mobile phase in chromatography, for example, methanol, acetonitrile, water, ethanol, 1-propanol, 2-propanol, acetone, tetrahydrofuran (THF), and mixed solvents thereof. A solvent that can be freely mixed with water is preferred. For example, a mixed solvent of acetonitrile, methanol, or water is preferred. Specifically, the solvent composition of the mobile phase can be determined, for example, by the ratio of water to acetonitrile (usually by volume), the ratio of water to methanol (usually by volume), etc.

[0018] (Step S02: Define the stationary phase model) In step S02, multiple stationary phase models representing the composition of the stationary phase are defined. In the retention time prediction method according to this embodiment, the actual stationary phase is represented as a superposition of multiple stationary phase models. For this reason, in step S02, it is preferable to define multiple stationary phase models with various compositional variations so that the interaction with the mobile phase and the sample to be analyzed by HPLC is sufficiently reflected.

[0019] (Step S03: Create a parametric function) In step S03, a parametric function is constructed using a linear combination of the partition coefficients between the mobile phase and the solvent composition determined in step S01 for the multiple stationary phase models defined in step S02. The linear combination used in the parametric function is constructed by summing up terms obtained by multiplying the partition coefficient of each stationary phase model by a parameter for all stationary phase models. The linear combination used in the parametric function represents the superposition of the multiple stationary phase models defined in step S02.

[0020] (Step S04: Obtain retention time and molecular structure information of the standard substance) In step S04, retention time and molecular structure information of multiple standard substances are obtained. The multiple standard substances can be any substances whose retention time and molecular structure are known. The molecular structure information can be any information that allows for the identification of the molecular structure, and the method of representing the molecular structure is not limited to a specific method.

[0021] (Step S05: Calculate the partition coefficient of the standard substance) In step S05, the partition coefficients between multiple standard substances and the mobile phase in multiple stationary phase models are calculated using quantum chemical calculations based on the molecular structure information obtained in step S04. Any quantum chemical calculation software can be used for these calculations.

[0022] (Step S06: Determine the parameters of the parametric function) In step S06, the parameters of each partition coefficient in the linear combination of the parametric functions prepared in step S03 are determined by linear regression using the retention times and partition coefficients of multiple standard materials. For linear regression, for example, the least squares method can be used.

[0023] (Step S07: Create a specific function) In step S07, a specific function is created by substituting the parameters determined in step S06 into the parametric function created in step S03.

[0024] (Step S08: Obtain molecular structure information of the target substance) Step S08 involves obtaining molecular structure information of the target substance. The target substance is the substance whose retention time is to be predicted, and is, for example, a substance expected to be contained in a sample to be analyzed by HPLC. The target substance can be arbitrarily selected from substances whose molecular structure is known. The molecular structure information only needs to be information that can identify the molecular structure, and the method of representing the molecular structure is not limited to a specific method.

[0025] (Step S09: Calculate the partition coefficient of the target substance) In step S09, the partition coefficients between the target substance and the mobile phase in multiple stationary phase models are calculated using quantum chemical calculations based on molecular structure information. The same quantum chemical calculation software used in step S05 can be used for the quantum chemical calculations.

[0026] (Step S10: Calculate the retention time of the target substance) In step S10, the retention time is calculated for the target substance from the partition coefficient calculated in step S09, using the specific function created in step S07.

[0027] (Other configurations) Steps S01 to S10 do not necessarily have to be performed in this order. For example, steps S08 and S09 for the target substance may be performed before steps S04 and S05 for the standard substance.

[0028] [Specific example] (Isocratic elution) A specific example of a method for predicting retention time obtained by HPLC using isocratic elution will be described based on the configuration of the above embodiment. First, in order to demonstrate the advantages of the retention time prediction method according to this embodiment, a general method for predicting retention time obtained by HPLC using isocratic elution will be described.

[0029] Figure 2 is a schematic model showing the mobile and stationary phases in a column during HPLC using isocratic elution. This model assumes that the concentrations of the target substance in both the mobile and stationary phases reach partition equilibrium instantaneously. As shown in Figure 2, the concentration of the target substance in the stationary phase is C s The concentration of the target substance in the mobile phase is set to C m Therefore, the distribution coefficient P is expressed by the following equation (1).

[0030]

number

[0031] Here, assuming that only the molecules of the target substance that are in the mobile phase move, let u be the velocity of the solvent in the mobile phase, and let V be the volume of the stationary phase. s Let V be the volume of the mobile phase. m Therefore, the migration velocity v of the target substance molecules in the mobile phase is expressed by the following equation (2).

[0032]

number

[0033] Therefore, if the column length is L, and the time required for the non-retained solvent in the mobile phase to pass through the column is t0, then the retention time t from the injection of the target substance into the column until it reaches the end of the column is t. R This is expressed by equation (3) below.

[0034]

number

[0035] Therefore, if the partition coefficient P is calculated by quantum chemical calculation, the retention time t can be obtained from equation (3). R This allows for the calculation of the partition coefficient P. However, software suitable for quantum chemical calculations of molecules being analyzed does not accurately reflect the properties of silica gel packing materials, such as chemically bonded porous spherical silica gel whose surface is modified with octadecylsilyl groups, which constitute the stationary phase of ODS columns commonly used in HPLC. Therefore, it is difficult to obtain high accuracy in predicting retention time.

[0036] In contrast, the retention time prediction method according to this embodiment is configured to predict the retention time with high accuracy regardless of the stationary phase configuration. Specifically, in the retention time prediction method according to this embodiment, the retention time t R By using the parametric function shown in equation (4) below, which is expressed as a linear combination of the distribution coefficients between the stationary phase and the mobile phase in multiple stationary phase models, the retention time t R It can predict with high accuracy.

[0037]

Number

[0038] (Gradient elution) A method for predicting the retention time obtained by HPLC with gradient elution will be specifically described according to the configuration of the above embodiment.

[0039] FIG. 3 is a model schematically showing the mobile phase and stationary phase in a column in HPLC with gradient elution. In this model, the initial state (t = 0) and the state at t = t are shown. Also, it is assumed that the position of the molecules of the target substance moves from x = 0 to x = x from t = 0 to t = t. As shown in FIG. 3, the concentration of the molecules of the target substance in the stationary phase at time t is C s (t), and the concentration of the molecules of the target substance in the mobile phase at time t is C m (t). Then, the partition coefficient P is expressed by the following formula (5).

[0040]

Number

[0041] Figure 4 is a schematic model showing only the mobile phase in a column during HPLC with gradient elution. This model assumes that the influence of changes in the mobile phase composition on the stationary phase is sufficiently small, and ignores changes in the stationary phase over time. This model shows the initial (t=0) state and the t=t state of the mobile phase. This model assumes that the initial (t=0) mobile phase and the t=t mobile phase remain separated without mixing, and that the target substance is in partition equilibrium between them. Quantum chemical calculations are possible even for such a model. Using such a model, the time-dependent partition coefficient can be converted to a value at a specific time and evaluated. For example, converting to the initial (t=0) value allows for evaluation of the partition coefficient under the same conditions without considering the influence of changes in the mobile phase composition, thus obtaining high accuracy. As shown in Figure 4, the concentration of the target substance molecules in the mobile phase at the initial (t=0) state is c0, and the concentration of the target substance molecules in the mobile phase at t=t state c t Therefore, the distribution coefficient p(t) between the initial (t=0) mobile phase and the mobile phase at t=t is expressed by the following equation (6).

[0042]

number

[0043] Here, if N is the total number of molecules of the target substance in the stationary phase and mobile phase, then the concentration C m (t) is expressed by the following equation (7), and the concentration C s (t) is expressed by the following equation (8).

[0044]

number

[0045] Furthermore, if the volume of the stationary phase is Vs and the volume of the mobile phase is Vm, then the following relationships (9), (10), and (11) hold.

[0046]

number

[0047] Let u be the velocity of the solvent in the mobile phase, v(t) be the velocity of the target substance molecules in the mobile phase, L be the length of the column, t0 be the time required for the non-retained solvent to pass through the column, and t be the retention time for the target substance from injection to the end of the column. R Therefore, the following equation (12) holds true.

[0048]

number

[0049] Equation (12) can be transformed into equation (13) below.

[0050]

number

[0051] Furthermore, the distribution coefficient P(t) can be expressed using equation (11) and P(0) and p(t) by the following equation (14).

[0052]

number

[0053] Substituting equation (14) into equation (13), we obtain equation (15) below.

[0054]

number

[0055] In equation (15), (Vs / Vm)P(0) represents the ratio of the number of molecules of the target substance in the stationary phase and the mobile phase at the solvent composition of the mobile phase at t=0, and is called the retention coefficient. In the retention time prediction method according to this embodiment, the retention coefficient (Vs / Vm)P(0) is expressed as a parametric function shown in equation (16) below, which is a linear combination of the partition coefficients between the mobile phase and multiple stationary phase models, and the retention time t R It can predict with high accuracy.

[0056]

number

[0057] (Other configurations) The retention time prediction method according to this embodiment is not limited to the above configuration. For example, in the above description of HPLC by gradient elution, the model shown in Figure 3 was used to convert the state at any time t to the initial (t=0) state, but it is not limited to this, and the configuration may be used to convert the state at any time t to a specific time t other than the initial (t=0) state. Furthermore, to predict the retention time obtained by HPLC by isocratic elution, a parametric function in which the retention coefficient is expressed as a linear combination of partition coefficients may be used. Moreover, to predict the retention time obtained by HPLC by gradient elution, the retention coefficient for the mobile phase composition at a specific time may be converted to retention time, and a parametric function in which that retention time is expressed as a linear combination of partition coefficients may be used.

[0058] [Preparative chromatography] By using the retention time prediction method according to this embodiment, it becomes possible to accurately separate the components contained in a sample by preparative chromatography. Specifically, first, the separation conditions for each component contained in the sample are explored using the retention time prediction method according to this embodiment. In order to optimize the separation conditions, it is effective to consider the peak width and peak height of each component that appears in the chromatogram, in addition to the retention time. For example, the separation conditions for separating the main component and the minor component of interest contained in the sample, or the separation conditions for separating the minor components from each other, are considered to need to be set so that their peaks are sufficiently separated. To this end, for example, optimizing the eluent composition of the mobile phase is effective, and the retention time prediction method according to this embodiment can be used to determine whether sufficient differences in retention times are ensured for the separation of each component. Then, using the separation conditions obtained through the exploration, the components contained in the sample are separated by preparative chromatography.

[0059] [Method for identifying components contained in a sample] By using the retention time prediction method according to this embodiment, the components contained in a sample can be accurately identified. Specifically, first, a chromatogram of the sample is prepared by HPLC. Then, using the retention time calculated by the prediction method according to this embodiment, the components contained in the sample are identified from the chromatogram prepared by HPLC.

[0060] Furthermore, the identification method according to this embodiment allows for more accurate identification of components contained in a sample by using information other than the retention time calculated by the prediction method according to this embodiment. For example, by using the retention time calculated by the prediction method according to this embodiment to search for separation conditions for components contained in a sample, and then analyzing the components separated from the sample by preparative chromatography using the separation conditions obtained through the search, it becomes possible to more accurately identify those components.

[0061] For example, the ultraviolet-visible absorption spectrum of components separated from a sample by preparative chromatography can be measured. In this case, by comparing the absorption wavelength and / or absorption intensity obtained by measurement with the absorption wavelength and / or absorption intensity of the compound predicted by computational chemistry, and considering their similarity, the components separated from the sample can be identified more accurately.

[0062] Instruments equipped with detectors capable of detecting ultraviolet-visible absorption spectra, such as spectrophotometers, can use these measurement results to more accurately identify the components contained in a sample, similar to the method used by preparative chromatography. For example, in a device equipped with a detector capable of detecting ultraviolet-visible absorption spectra, The components contained in the sample are separated using the retention time calculated by the prediction method according to this embodiment. The ultraviolet-visible absorption spectra of the separated components were measured, By comparing the absorption wavelength and / or absorption intensity obtained by the above-mentioned measurement with the absorption wavelength and / or absorption intensity of the compound predicted from computational chemistry, the component can be identified by considering its similarity.

[0063] Furthermore, mass spectrometry of the components separated from the sample is also effective. In this case, by considering the molecular weight obtained by mass spectrometry, the components separated from the sample can be identified more accurately.

[0064] With a mass spectrometer, the measurement results can be used directly to identify the components of a sample more accurately, similar to the method used by preparative chromatography. For example, in a device equipped with a mass spectrometer, The components contained in the sample are separated using the retention time calculated by the prediction method according to this embodiment. The components can be identified by considering the molecular weight obtained by the mass spectrometry.

[0065] [Judgment method] The identification method according to this embodiment can be used to determine the presence or absence of sensitizing substances and / or toxic substances in a sample. Specifically, first, information on sensitizing substances and / or toxic substances is obtained. Sensitizing substances and / or toxic substances information is typically a database listing sensitizing substances and / or toxic substances. Specifically, examples of sensitizing substances and / or toxic substances information include the "Integrated Platform for Toxic Assessment Support Systems (HESS)" and the "Comprehensive Chemical Substances Information Provision System (CHRIP)" of the National Institute of Technology and Evaluation. Then, by comparing the information on the components contained in the sample identified by the identification method according to this embodiment with the sensitizing substances and / or toxic substances information, the presence or absence of sensitizing substances and / or toxic substances in the sample can be determined.

[0066] [Estimation method] The identification method according to this embodiment can be used to estimate the relative concentrations of multiple components contained in a sample. For example, in HPLC using a detector that can acquire intensity information correlated with the amount of each component contained in the sample, the relative concentrations of multiple components contained in the sample can be estimated using the intensity information obtained by the detector. The intensity information preferably includes, for example, the absorbance in the ultraviolet-visible region.

[0067] Specifically, in the estimation method according to this embodiment, first intensity information correlated with the amounts of multiple components is calculated from quantum chemical calculations based on information about multiple components contained in the sample identified by the identification method according to this embodiment. Furthermore, second intensity information of the same type as the first intensity information, also correlated with the amounts of multiple components, is obtained using an HPLC detector. Then, the relative concentrations of the multiple components in the sample can be estimated using the first and second intensity information.

[0068] [Methods for predicting bioaccumulation] In this embodiment, the bioaccumulation potential of a target substance can be predicted in a manner similar to the prediction of retention time obtained by HPLC. In the bioaccumulation prediction method according to this embodiment, the mobile phase is water, and multiple stationary phase models are defined. Next, a parametric function is created using a linear combination of the partition coefficients between the target substance and water in the multiple stationary phase models. Next, bioaccumulation potential and molecular structure information of multiple standard substances are obtained. Next, the partition coefficients between the target substance and water in the multiple stationary phase models are calculated using quantum chemical calculations based on the molecular structure information of the multiple standard substances. Next, the parameters of the partition coefficients in the linear combination of the parametric function are determined by linear regression using the bioaccumulation potential and partition coefficients of the multiple standard substances. Next, a specific function is created by substituting the parameters into the parametric function. Next, molecular structure information of the target substance is obtained. Next, the partition coefficients between the target substance and water in the multiple stationary phase models are calculated using quantum chemical calculations based on the molecular structure information of the target substance. Finally, the bioaccumulation potential of the target substance is calculated from the partition coefficients using the specific function.

[0069] [Other embodiments] By using the retention time calculated by the prediction method according to this embodiment, various other possibilities become available.

[0070] For example, by using the retention time calculated by the prediction method according to this embodiment, it is possible to characterize the components and composition contained in a sample, and furthermore, to characterize phenomena from the components and composition. In addition, by using the retention time calculated by the prediction method according to this embodiment, quantitative and qualitative analysis of the components and composition contained in a sample becomes possible, and high purity is achieved. Furthermore, by using the retention time calculated by the prediction method according to this embodiment, efficient analysis becomes possible, and analysis becomes easy even for inexperienced individuals.

[0071] Furthermore, in the preparative chromatography according to this embodiment, by using the retention time calculated by the prediction method according to this embodiment, it is possible to explore analytical conditions that achieve both a reduction in analysis time and good separation.

[0072] Furthermore, by using the retention time calculated by the prediction method according to this embodiment, it is possible to predict the chromatogram obtained by HPLC for samples with known main and minor components without actually performing HPLC. In this case, in HPLC using an ultraviolet-visible detector, the chromatogram can be predicted more accurately by estimating the absorbance from quantum chemical calculations. The method for estimating the absorption wavelength, absorption intensity, and / or absorption spectrum in the ultraviolet-visible region by quantum chemical calculations is not limited, but for example, it can be estimated using time-dependent density functional theory (TDDFT), or it can be calculated using Dassault Systems' TURBOMOLE ver.7-5-1,2021 (for example, with the functional b3-lyp and the basis set def-TZVP).

[0073] In addition, by using the retention time calculated by the prediction method according to this embodiment, the relative concentration to the main component can be estimated by estimating the absorption intensity from quantum chemical calculations in HPLC using a UV-Vis detector as the detector. In this case, when multiple components are selected from a sample, the composition ratio (usually the mass ratio) of those multiple components can be estimated.

[0074] [Examples] The following describes embodiments of the present invention, but these embodiments are merely examples of the present invention, and the configuration of the present invention is not limited to the configuration of these embodiments.

[0075] (Example 1) In Example 1, retention times obtained by HPLC using isocratic elution were predicted. The target substances for retention time prediction were the following target substances 1 to 28. All of target substances 1 to 28 are substances for which the measured retention times obtained by HPLC using isocratic elution are known. In Example 1, the accuracy of retention time prediction was evaluated by comparing the predicted retention times for target substances 1 to 28 with the measured values.

[0076] Target substance 1: α-terpineol Target substance 2: Benzyl acetate Target substance 3: Biomugueto Target substance 4: cis-3-hexen-1-ol Target substance 5: cis-3-hexenyl acetate Target substance 6: cis-3-hexenyl salicylate Target substance 7: cis-jasmon Target substance 8: Helional Target substance 9: Hexyl salicylate Target substance 10: Indole Target substance 11: Methyl dihydrojasmonate Target substance 12: Methylisoeugenol Target substance 13: Methylpamplemousse Target substance 14: 3-methyl-5-phenyl-1-pentanol Target substance 15: Sandal Mysore core Target substance 16: 1-phenylethyl acetate Target substance 17: Dodecanal Target substance 18: Allyl heptanoate Target substance 19: Chalon Target substance 20: Citronellyl acetate Target substance 21: Citronellyl formate Target substance 22: Damascenone Target substance 23: β-ionone Target substance 24: l-citronellol Target substance 25: Magantol Target substance 26: Methyl anthranilate Target substance 27: 2-methyl noninate Target substance 28: Allyl cyclohexanepropionate

[0077] The mobile phase solvent composition (volume ratio) was set to four different values: acetonitrile:water = 100:0, acetonitrile:water = 90:10, acetonitrile:water = 80:20, and acetonitrile:water = 60:40. Phosphoric acid was added to the mobile phase solvent to a concentration of 0.1 w / v%. The column temperature was 40°C. The column used was an L-column ODS (particle size 5 μm, 4.6 × 100 mm) from the Chemicals Evaluation and Research Institute.

[0078] As stationary phase models, we defined stationary phase models 1 to 25, which vary the ratio of acetonitrile, water, and octadecane, and stationary phase models 26 to 30, which have aromatic rings, taking into account the interaction with the target substance itself when the target substance has an aromatic ring. Table 1 shows the composition of stationary phase models 1 to 25. Table 1 shows the mass ratio (mass %) of acetonitrile, water, and octadecane for stationary phase models 1 to 25. Table 2 shows the composition of stationary phase models 26 to 30. Stationary phase models 26 to 30 consist only of the substances shown in Table 2.

[0079] [Table 1]

[0080] [Table 2]

[0081] The reference materials used were the following reference materials 1 to 156.

[0082] Standard substance 1: Uracil Standard substance 2: Benzenesulfonic acid Standard substance 3: Linalool Standard substance 4: (R)-(+)-limonene Standard substance 5: Geraniol Standard substance 6: Citronellol Standard substance 7: Benzyl alcohol Standard substance 8: Benzyl benzoate Standard substance 9: Eugenol Standard substance 10: Methyl eugenol Standard substance 11: Isoeugenol Standard substance 12: Cinnamyl alcohol Standard substance 13: α-hexyl cinnamaldehyde Standard substance 14: Benzyl salicylate Standard substance 15: (-)-Perillaldehyde Standard substance 16: Hydroxycitronellal Standard substance 17: Coumarin Standard substance 18: 4-Methoxybenzyl alcohol Standard substance 19: Cinnamaldehyde Standard substance 20: Cinnamain Standard substance 21: Caffeine Standard substance 22: Pentylbenzene Standard substance 23: Butylbenzene Standard substance 24: Methyl benzoate Standard substance 25: α-Amyl cinnamaldehyde Standard substance 26: Lilial Standard substance 27: 2-methyl octinate Standard substance 28: α-Damascone Standard substance 29: o-terphenyl Standard substance 30: Camphor Standard substance 31: Ethyl benzoate Standard substance 32: tert-butylhydroquinone Standard substance 33: Purpurine Standard substance 34: 1,4-dihydroxyanthraquinone Standard substance 35: Potassium sorbate Standard substance 36: 2-Phenoxyethanol Standard substance 37: Guaiazulene Standard substance 38: Acetanilide Standard substance 39: Propiophenone Standard substance 40: Benzophenone Standard substance 41: Acetophenone Standard substance 42: Hexanophenone Standard substance 43: Valerophenone Standard substance 44: Butyrophenone Standard substance 45: Naphthalene Standard substance 46: Benz[a]anthracene Standard substance 47: Benzo[a]pyrene Standard material 48: Anthracene Standard substance 49: Benzo[b]fluorantene Standard substance 50: Chrysene Standard substance 51: Triphenylene Standard substance 52: Dibenz[a,h]anthracene Standard substance 53: N-(3-(dimethylamino)propyl)acrylamide Standard substance 54: N-tert-butylacrylamide Standard substance 55: N,N-dimethylacrylamide Standard substance 56: 4-Isopropylphenol Standard substance 57: Dibutylhydroxytoluene Standard substance 58: Octanophenone Standard substance 59: Curcumin Standard substance 60: Bisdemethoxycurcumin Standard substance 61: Demethoxycurcumin Standard substance 62: Fluorantene Standard substance 63: (-)-epicatechin gallate Standard substance 64: Epigallocatechin Standard substance 65: Cyanidanol Standard substance 66: α-Amyl cinnamyl alcohol Standard substance 67: Butyloctyl salicylate Standard substance 68: 4-dimethylaminocinnamaldehyde Standard substance 69: Surisobenzone Standard substance 70: Triphenylmethane Standard substance 71: Homosalate Standard substance 72: 1-Methoxy-4-methylbenzene Standard substance 73: o-xylene Standard substance 74: m-xylene Standard substance 75: p-xylene Standard substance 76: Phenol Standard substance 77: Tonalide Standard substance 78: Ethylhexyl triazone Standard substance 79: Benzene Standard substance 80: Benzonitrile Standard substance 81: Biphenyl Standard substance 82: 4-Chlorobenzaldehyde Standard substance 83: 4-chlorophenol Standard substance 84: o-cresol Standard substance 85: 1,2-dichlorobenzene Standard substance 86: Diethyl phthalate Standard substance 87: Dimethyl phthalate Standard substance 88: 1,3-dinitrobenzene Standard substance 89: 1,2-dinitrobenzene Standard substance 90: Diphenyl ether Standard substance 91: Ethylbenzene Standard substance 92: 1-fluoro-3-nitrobenzene Standard substance 93: 1-fluoro-2-nitrobenzene Standard substance 94: 1-fluoro-4-nitrobenzene Standard substance 95: 4-fluorophenol Standard substance 96: 4-hydroxybenzaldehyde Standard substance 97: 4-Methoxybenzaldehyde Standard substance 98: 4-methylbenzaldehyde Standard substance 99: 4'-nitroacetophenone Standard substance 100: 4-nitrobenzaldehyde Standard substance 101: Bromobenzene Standard substance 102: Nitrobenzene Standard substance 103: 4-phenylphenol Standard substance 104: 3-phenyl-1-propanol Standard substance 105: Propylbenzene Standard substance 106: 2,6-dichlorobenzyl chloride Standard substance 107: 2,4-dimethylphenol Standard substance 108: 2-phenylethanol Standard substance 109: Aniline Standard substance 110: m-cresol Standard substance 111: 4-Chlorobenzenesulfonamide Standard substance 112: p-cresol Standard substance 113: Methylparaben Standard substance 114: Benzamide Standard substance 115: Benzyl bromide Standard substance 116: Phenylacetonitrile Standard substance 117: Salicylic acid Standard substance 118: Chlorphenesin Standard substance 119: Ethylparaben Standard substance 120: Dimethoxydi-p-cresol Standard substance 121: Disodium ascorbic acid sulfate Standard substance 122: 4'-methoxyacetophenone Standard substance 123: 3-hydroxybenzoic acid Standard substance 124: Acenaphthene Standard substance 125: Amitriptyline hydrochloride Standard substance 126: 2,2'-[[4-[(2-chloro-4-nitrophenyl)azo]-3-methylphenyl]imino]bisethanol Standard substance 127: Solvent green 3 Standard substance 128: 1,4-dinitrobenzene Standard substance 129: Bis(2-ethylhexyl) isophthalate Standard substance 130: Melamine monomer Standard substance 131: N-lauroyl sarcosine Standard substance 132: 3-Ethoxy-4-hydroxybenzaldehyde Standard substance 133: 4-hydroxy-6-methyl-2-pyrone Standard substance 134: Toluene Standard substance 135: Styrene Standard material 136: Indene Standard substance 137: Fluorobenzene Standard substance 138: Nicotinamide Standard substance 139: 4-Isopropyl-3-methylphenol Standard substance 140: Ammonium glycyrrhizinate Standard substance 141: Vitamin E nicotinate ester Standard substance 142: Allantoin Standard substance 143: Levocarnitine chloride Standard substance 144: Dexpanthenol Standard substance 145: 3,4-dimethyl-5-phenyloxazolidine Standard substance 146: Benzyl nicotinate Standard substance 147: Oxybenzone Standard substance 148: Diethylamino hydroxybenzoyl hexyl benzoate Standard substance 149: Avobenzone Standard substance 150: Octinoxate Standard substance 151: Octisalate Standard substance 152: Bisoctrizole Standard substance 153: Benzoic acid Standard substance 154: Benzoic anhydride Standard substance 155: Maleic acid Standard substance 156: Fumaric acid

[0083] For standard substances 1-156, the partition coefficients between the mobile phase and stationary phase models 1-30 were calculated using quantum chemical calculations based on molecular structure information. Quantum chemical calculations were performed using Dassault Systems' TURBOMOLE ver.7-5-1,2021 and COSMOtherm ver.21.0,2021, at TZVPD-Fine level and column temperature of 40°C. When standard substances and target substances could exist in several states due to dissociation equilibrium reactions, the structure of the state stable under acidic conditions was used in the calculations, taking the measurement conditions into consideration.

[0084] Using the retention times and partition coefficients of standard substances 1 to 156, the parameters of the partition coefficients in the linear combination of the parametric functions shown in equation (4) above were determined by the least squares method, and a specific function was constructed by substituting these parameters into the parametric functions shown in equation (4) above.

[0085] Next, for target substances 1-28, the partition coefficients between the mobile phase and stationary phase models 1-30 were calculated using quantum chemical calculations based on molecular structure information, in the same manner as for standard substances 1-156 described above. Then, for target substances 1-28, predicted retention times were calculated from the partition coefficients using a specific function. Similarly, for standard substances 1-156, predicted retention times were also calculated from the partition coefficients using a specific function.

[0086] Figures 5-8 show the holding time t on the horizontal axis. R Let the predicted value be the holding time t on the vertical axis. R The graph shows the measured values ​​for target substances 1-28 and standard substances 1-156, each represented by a single plot. Figure 5 shows the results obtained when the solvent composition (volume ratio) of the mobile phase is acetonitrile:water = 100:0. Figure 6 shows the results obtained when the solvent composition (volume ratio) of the mobile phase is acetonitrile:water = 90:10. Figure 7 shows the results obtained when the solvent composition (volume ratio) of the mobile phase is acetonitrile:water = 80:20. Figure 8 shows the results obtained when the solvent composition (volume ratio) of the mobile phase is acetonitrile:water = 60:40.

[0087] Referring to Figures 5-8, it can be seen that in all configurations of the mobile phase solvent composition, the plots for target substances 1-28 and standard substances 1-156 lie on a straight line with a slope of approximately 1, indicating a direct proportionality, and predictive values ​​close to the measured values ​​for retention time were obtained.

[0088] (Example 2) In Example 2, the retention time obtained by HPLC using gradient elution was predicted. The target substance for retention time prediction was 5HIAA (5-hydroxyindoleacetic acid), an indole derivative. 5HIAA is a substance for which the measured retention time obtained by HPLC using gradient elution is known. In Example 2, the accuracy of the retention time prediction was evaluated by comparing the predicted retention time with the measured retention time for the target substance.

[0089] [ka]

[0090] The column used for the measurements was an L-column ODS (particle size 5 μm, 2.1 × 150 mm) from the Chemicals Evaluation and Research Institute, with a column temperature of 40°C. Formic acid was added to the mobile phase solvent to a concentration of 0.1 v / v%. The initial solvent composition (volume ratio) of the mobile phase (t=0) was acetonitrile:water = 5:95. As stationary phase models, stationary phase models 31 to 34 were defined, each with a different ratio of water to octadecan. Table 3 shows the mass ratios (mass%) of water and octadecan for stationary phase models 31 to 34.

[0091] [Table 3]

[0092] The standard substances used were the following indole derivatives: IAA (indole-3-acetic acid), IGA (3-indole glyoxidic acid), MIG (3-indole glyoxylate methyl), 5HTP (5-hydroxytryptophan), and Trp (tryptophan).

[0093] [ka]

[0094] For standard substances, the partition coefficients between the mobile phase and stationary phase in stationary phase models 31-34 were calculated using quantum chemical calculations based on molecular structure information. Quantum chemical calculations were performed using Dassault Systems' TURBOMOLE ver.7-5-1,2021 and COSMOtherm ver.21.0,2021, at TZVPD-Fine level and column temperature of 40°C.

[0095] Furthermore, for the standard substance, p(t) in equation (15) above was calculated by quantum chemical calculation. Figure 9 shows the volume fraction φ of acetonitrile in the mobile phase, which is the measurement condition. AThis shows the time evolution. The time t0 for the unretained solvent in the mobile phase to pass through the column is 1.60 minutes. If x(t) is the position of the target substance molecule at time t, then φ is the volume fraction of acetonitrile at position x(t) of the target substance molecule. A (t) is expressed by the following equation (17).

[0096]

number

[0097] Volume fraction φ of acetonitrile in the mobile phase A As a function of logp(φ shown in equation (18) below, A (t)) was calculated using COSMOtherm at TZVPD-Fine level and column temperature of 40°C.

[0098]

number

[0099] Figure 10 shows logp(φ) for each standard substance. A The calculation result for (t)) is shown. Approximating the curve shown in Figure 10 with a quadratic equation yields the following equation (19).

[0100]

number

[0101] Furthermore, for each standard substance, the retention time t in the following equation (20) R The retention coefficient (Vs / Vm)P(0) was calculated by quantum chemical calculation using the following equations (21), (22), (23), and (24) so ​​that it corresponds to the known retention time of each standard substance.

[0102]

number

[0103]

number

[0104] Figure 11 shows the results of fitting using quantum chemical calculations. In Figure 11, the horizontal axis represents time, and the vertical axis represents the position of the standard substance molecules in the mobile phase. From these results, the retention coefficient (Vs / Vm)P(0) for each standard substance was determined.

[0105] Using the (Vs / Vm)P(0), p(t), and partition coefficients of each standard substance, the parameters of the partition coefficients in the linear combination of the parametric functions shown in equation (16) above were determined by the least squares method, and a specific function was constructed by substituting these parameters into the parametric function shown in equation (16) above.

[0106] For the target substance, the partition coefficients between the mobile phase and the stationary phase in stationary phase models 31-34 were calculated by quantum chemical calculations using molecular structure information, in the same manner as for the standard substance described above. Then, for the target substance, the retention coefficient (Vs / Vm)P(0) was calculated from the partition coefficient using a specific function. The retention coefficient (Vs / Vm)P(0) for the target substance was 6.04. Furthermore, for the target substance, the retention time t was calculated from the retention coefficient (Vs / Vm)P(0) using equation (15) above. R The predicted value was calculated for the retention time t of the target substance. R The predicted value was 9.34 minutes, which is close to the actual measured value of 9.43 minutes.

Claims

1. A method for predicting the retention time of a target substance obtained by high-performance liquid chromatography, Determine the solvent composition of the mobile phase. Define multiple stationary phase models, A parametric function is constructed using a linear combination of the distribution coefficients between the mobile phase and the multiple stationary phase models. We obtained retention time and molecular structure information for multiple standard substances. For the aforementioned multiple standard substances, the partition coefficients between the mobile phase and the multiple stationary phase models are calculated by quantum chemical calculations using molecular structure information. Using the retention times and partition coefficients of the aforementioned multiple standard substances, the parameters of the partition coefficients in the linear combination of the parametric function are determined by linear regression. A specific function is created by substituting the parameters into the aforementioned parametric function. The molecular structure information of the aforementioned target substance is obtained, Using molecular structure information, the partition coefficients between the target substance and the mobile phase in the multiple stationary phase models are calculated by quantum chemical calculations. Using the aforementioned specific function, the retention time is calculated from the partition coefficient for the target substance. Method for predicting retention time.

2. In the parametric function, the retention time is expressed by a linear combination of the distribution coefficients between the mobile phase and the multiple stationary phase models. A method for predicting retention time according to claim 1.

3. In the parametric function, the retention coefficient is expressed as a linear combination of the distribution coefficients between the mobile phase and the multiple stationary phase models. For the aforementioned multiple standard substances, the retention coefficient is calculated from the retention time and molecular structure information. To determine the aforementioned parameters, the retention coefficients and partition coefficients of the plurality of standard materials are used, Using the aforementioned specific function, the retention coefficient is calculated from the distribution coefficient for the target substance. For the aforementioned target substance, the retention time is calculated from the retention coefficient. A method for predicting retention time according to claim 1.

4. A method for predicting the retention time of a target substance obtained by high-performance liquid chromatography by gradient elution, The retention coefficient in the parametric function is the retention coefficient of the mobile phase at a specific time in the solvent composition. The method for predicting retention time according to claim 3.

5. The retention coefficient in the parametric function is the retention coefficient in the initial solvent composition of the mobile phase. The method for predicting retention time according to claim 4.

6. A method for identifying components contained in a sample, A chromatogram of the sample is prepared by high-performance liquid chromatography. Using the retention time calculated by the prediction method described in any one of claims 1 to 5, the components contained in the sample are identified from the chromatogram. A method for identifying the components contained in a sample.

7. A preparative chromatography method for separating components contained in a sample, Using the retention time calculated by the prediction method described in any one of claims 1 to 5, the separation conditions for the components contained in the sample are explored. The components contained in the sample are separated using the separation conditions obtained by the above exploration. Preparative chromatography.

8. Using the retention time calculated by the prediction method described in any one of claims 1 to 5, the separation conditions for the components contained in the sample are explored. The components contained in the sample are separated using the separation conditions obtained by the above exploration. The ultraviolet-visible absorption spectra of the separated components were measured, The absorption wavelength and / or absorption intensity obtained by the above-mentioned measurement are compared with the absorption wavelength and / or absorption intensity of the compound predicted from computational chemistry, and the component is identified by considering the similarity between them. The identification method according to claim 6.

9. Using the retention time calculated by the prediction method described in any one of claims 1 to 5, the separation conditions for the components contained in the sample are explored. The components contained in the sample are separated using the separation conditions obtained by the above exploration. Mass spectrometry was performed on the separated components. The components are identified considering the molecular weight obtained by the mass spectrometry. The identification method according to claim 6.

10. Obtain information on sensitizing substances and / or toxic substances, The presence or absence of sensitizing substances and / or toxic substances in the sample is determined by comparing the information of the components contained in the sample identified by the identification method described in claim 6 with the information of the sensitizing substances and / or toxic substances. Judgment method.

11. A high-performance liquid chromatography system using a detector capable of obtaining intensity information correlated with the amount of a component, Based on the information of the plurality of components contained in the sample identified by the identification method described in claim 6, first intensity information of the plurality of components is calculated from quantum chemical calculations. The detector acquires second intensity information of the plurality of components, The relative concentrations of the plurality of components in the sample are estimated using the first intensity information and the second intensity information. Estimation method.

12. The aforementioned intensity information includes the absorption intensity in the ultraviolet-visible region. The estimation method according to claim 11.

13. A method for predicting the bioaccumulation of a target substance, Define multiple stationary phase models, A parametric function is constructed using a linear combination of the distribution coefficients between the multiple stationary phase models and water. We obtained bioaccumulation and molecular structure information for multiple standard substances. For the aforementioned multiple standard substances, the partition coefficients between them and water in the aforementioned multiple stationary phase models are calculated by quantum chemical calculations using molecular structure information. Using the bioaccumulation properties and distribution coefficients of the aforementioned multiple standard substances, the parameters of the distribution coefficients in the linear combination of the parametric function are determined by linear regression. A specific function is created by substituting the parameters into the aforementioned parametric function. The molecular structure information of the aforementioned target substance is obtained, Using molecular structure information, the partition coefficient between the target substance and water in the multiple stationary phase models is calculated by quantum chemical calculations. Using the aforementioned specific function, the bioaccumulation potential of the target substance is calculated from its distribution coefficient. Methods for predicting bioaccumulation.

Images