Method for evaluating the composition

Abstract

The present invention addresses the issue of providing a composition evaluation method that is simple and enables the state of a composition to be understood, without dismantling a target substance. This composition evaluation method includes: a step in which a composition that includes at least one type of target substance is mixed with a labeling substance that has an interactive section and a fluorescent light-emitting section, to prepare a mixture; a step in which a fluorescence fingerprint is obtained for the mixture; and a step in which the composition state is evaluated on the basis of the fluorescence fingerprint. The interactive section has: a nucleic acid structure that can interact with each of a plurality of types of target substances; and / or a nucleic acid structure that can interact at a plurality of positions on each of the at least one type of target substance.

Description

[Technical Field]

[0001] The present invention relates to a method for evaluating a composition. [Background technology]

[0002] In recent years, amid the COVID-19 virus infection outbreak, a virtually unprecedented mRNA vaccine has been developed and has been shown to have a higher efficacy rate than conventional vaccines. This development has led to the development of mRNA formulations for various applications, and further advances are expected. However, mRNA is generally unstable and is easily degraded by RNase, an RNA-degrading enzyme present in the air. Therefore, attempts have been made to protect mRNA vaccines using lipid nanoparticles and other materials. However, long-term storage and environmental changes during transportation can lead to aggregation of lipid nanoparticles and leakage of the encapsulated mRNA, leading to degradation. Therefore, storing and transporting mRNA vaccines can be difficult. Given these characteristics of mRNA, a simple test that can determine vaccine readiness and calculate the degree of degradation immediately before use is expected to be necessary for quality assurance purposes.

[0003] Generally, RNA is verified by real-time PCR or capillary gel electrophoresis, but this requires purification of the RNA, and there are concerns about degradation during handling.

[0004] Meanwhile, in the clinical field, a method for quantifying the amount of various analytes based on the signal intensity of fluorescent dyes is known. However, this evaluation method makes it extremely difficult to distinguish between multiple identifiers (fluorochromes) when their excitation wavelengths or maximum wavelengths overlap. Therefore, there is a limit to the number of identifiers (fluorochromes) that can be used at one time, making this method difficult to apply when multiple components are mixed in the analyte.

[0005] Fluorescence fingerprinting technology, which has previously been difficult to utilize due to excessive data, can now be effectively utilized as inductive data through the use of artificial intelligence (AI). Fluorescence fingerprinting (or excitation-emission matrix) measurement is a brute-force measurement that measures fluorescence intensity while changing both the wavelength of the excitation light and the wavelength of the observed fluorescence. This fluorescence fingerprinting technology is used for evaluating food quality, for example. For example, the quality of food can be evaluated by analyzing the various fluorescent signals emitted by each component naturally contained in the food.

[0006] In recent years, attempts have been made to apply this fluorescence fingerprinting technology to quantification based on the signal intensity of the above-mentioned fluorescent dyes. For example, Patent Documents 1 and 2 describe a method of modifying an object with a fluorescent labeling substance that acts specifically on only the object, and evaluating the modified object using a fluorescence fingerprint.

[0007] Meanwhile, various fluorescent labeling substances that act specifically on target substances have also been developed, such as oligonucleotide labeling substances that contain a fluorescent group between adjacent nucleotides (e.g., Patent Document 3). When these labeling substances interact with the target nucleic acid, they emit a detectable fluorescent signal. [Prior art documents] [Patent documents]

[0008] [Patent Document 1] Japanese Patent Publication No. 2020-201202 [Patent Document 2] Special Publication No. 2012-507710 [Patent Document 3] International Publication No. 2010 / 001902 Summary of the Invention [Problem to be solved by the invention]

[0009] Here, it is conceivable to apply the fluorescent fingerprinting technology described in Patent Documents 1 and 2 to the evaluation of the degree of degradation of mRNA vaccines. However, if a labeling substance that specifically interacts with the target substance is used, as in Patent Documents 1 and 2, the labeling substance does not interact with the degraded product. Therefore, while the presence or absence of the target substance can be confirmed, its degradation state cannot be evaluated. On the other hand, it is conceivable to separately create other labeling substances that specifically interact with the degraded product. However, since various decomposition products are typically produced by degradation, it is necessary to prepare multiple labeling substances corresponding to each degraded product. In other words, this preparation requires a great deal of cost and time, making it impractical. Furthermore, this method cannot detect components when the target substance changes into an unexpected structure.

[0010] An object of the present invention is to provide an evaluation method that can easily grasp the state of a composition containing one or more target substances without separating the target substances. [Means for solving the problem]

[0011] The present invention provides the following method for evaluating a composition. A method for evaluating a composition, comprising the steps of: mixing a composition containing one or more types of target substance and a labeled substance having an interaction portion and a fluorescent portion to prepare a mixture; obtaining a fluorescence fingerprint of the mixture; and evaluating the state of the composition based on the fluorescence fingerprint, wherein the interaction portion has a nucleic acid structure capable of interacting with each of multiple types of target substance, and / or a nucleic acid structure capable of interacting at multiple positions on each of the one or more types of target substance.

[0012] The present invention also provides the following method for evaluating a composition. a step of irradiating the mixture with excitation light of a predetermined wavelength and measuring the fluorescence emitted by the mixture; and a step of evaluating the state of the objects from the results of the measurement, wherein the interaction part of the labeled substance has a nucleic acid structure capable of interacting with each of a plurality of types of the objects, and / or has a nucleic acid structure capable of interacting at a plurality of positions on each of the objects; and the wavelength of the excitation light is a wavelength identified from a state estimation model created by preparing a plurality of samples in which the states of the composition are different from one another, mixing the labeled substance with each of the samples to prepare a plurality of standard mixtures, obtaining standard fluorescence fingerprints for each of the standard mixtures, and subjecting the state of the composition in each of the samples and each of the standard fluorescence fingerprints to machine learning. [Effects of the Invention]

[0013] According to the composition evaluation method of the present invention, the state of a composition containing one or more types of target substance can be grasped in a simple manner without separating the target substance. [Brief explanation of the drawings]

[0014] [Figure 1] 1A and 1B are graphs showing examples of fluorescence fingerprints. [Figure 2] FIG. 2 is a diagram illustrating a flow of the first embodiment. [Figure 3] FIG. 3 is a diagram showing a flow of a modified example of the first embodiment. [Figure 4] FIG. 4 is a diagram showing the flow of a method for preparing a state estimation model. [Figure 5] FIG. 5 is a diagram illustrating a flow of the second embodiment. [Figure 6] FIG. 6 is a diagram showing a flow for determining a predetermined wavelength to be used in the second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0015] The present invention will be described in detail below with reference to two embodiments, although the present invention is not limited to these embodiments.

[0016] 1. First embodiment The flow of the composition evaluation method of this embodiment is shown in Figure 2. The composition evaluation method of this embodiment includes the steps of: step S11 (hereinafter also referred to as the "mixture preparation step") of mixing a composition containing one or more types of target substance and a labeling substance having an interacting moiety and a fluorescent moiety to prepare a mixture; step S12 (hereinafter also referred to as the "fluorescence fingerprint acquisition step") of obtaining a fluorescence fingerprint of the mixture; and step S13 (hereinafter also referred to as the "evaluation step") of evaluating the state of the composition based on the fluorescence fingerprint.

[0017] In this embodiment, a substance that acts nonspecifically on a target substance is used as the labeling substance. As used herein, "acting nonspecifically" means that the labeling substance can act on not only one type of substance but also multiple substances, or can act on not only one position on a specific substance but multiple positions. In this embodiment, the interacting portion of the labeling substance has a nucleic acid structure that can interact with each of multiple types of targets, or has a nucleic acid structure that can interact with multiple positions on each target.

[0018] As mentioned above, conventional methods for analyzing fluorescent signals have used labeling substances that react specifically with only the substance to be measured. In contrast, the inventors' extensive research has revealed that using labeling substances that interact nonspecifically with the target substance makes it possible to evaluate not only the presence or absence of the target substance but also the state of the composition. For example, when a labeling substance interacts with a specific target substance, the molecular structure surrounding the fluorescent moiety of the labeling substance and the interaction state of the interaction moiety differ between when the labeling substance interacts with a specific target substance and when the labeling substance interacts with a component converted from the specific target substance. Therefore, the fluorescent wavelength and fluorescence intensity emitted by the fluorescent moiety change subtly. Furthermore, when a labeling substance can interact with multiple sites on the target substance, a change in the state of the target substance can cause the labeling substance to no longer interact with the site that has changed in state, or even if it does interact, the fluorescent wavelength and fluorescence intensity change subtly due to the difference in the molecular structure surrounding the site. Therefore, if these differences can be detected, the state of the composition can be evaluated.

[0019] However, it is difficult to identify these differences using conventional methods for analyzing general fluorescent signals. Therefore, in this embodiment, a fluorescent fingerprint is used, and the state of the composition is identified based on the fluorescent fingerprint. When acquiring the fluorescent fingerprint, comprehensive measurements are performed while changing the wavelength of the excitation light and the wavelength of the fluorescent light to be observed. Therefore, it is easy to detect the above changes, and it becomes possible to identify the state of the composition in detail.

[0020] In this embodiment, as shown in Figure 2, the mixture preparation step S11, the fluorescence fingerprint acquisition step S12, and the evaluation step S13 may each be performed once. Alternatively, the mixture preparation step S11 and the fluorescence fingerprint acquisition step S12 may be performed multiple times using multiple types of labeling substances, varying the types of labeling substances or the combinations of labeling substances, and the evaluation step S13 may be performed based on multiple fluorescence fingerprints. Performing the evaluation step S13 based on multiple fluorescence fingerprints is preferable because it allows for more accurate evaluation of the state of the composition. Each step will be described below.

[0021] (1)Mixture preparation process In the mixture preparation step S11 of this embodiment, a composition containing one or more types of target substances and a labeling substance are mixed to prepare a mixture.

[0022] The composition used in this step may contain one or more target substances whose presence, concentration, degree of degradation, etc. are to be identified by interacting with a labeling substance. The composition may contain components other than the target substances, such as a solvent, within the scope that does not impair the purpose and effect of this embodiment.

[0023] The type of target substance is not particularly limited as long as it has at least a partial structure that can interact with the interacting portion (nucleic acid structure) of the labeling substance. Examples of targets include nucleic acids such as RNA and DNA, proteins such as amyloid beta, tartaric acid and polyphenols in wine, mycotoxins, microorganisms, fatty acids, fibers (animal hair, etc.), etc. Examples of targets also include those in which some or all of the above components have undergone structural changes, as well as decomposition products and aggregates of the above components.

[0024] For example, when measuring the degree of degradation of an mRNA vaccine, the deteriorated vaccine (composition) contains target substances such as mRNA and mRNA degradation products.

[0025] On the other hand, the labeling substance mixed with the composition in this step only needs to have an interacting portion containing a nucleic acid structure capable of interacting with the target substance and a fluorescent portion capable of emitting light upon irradiation with excitation light, and may also contain other structures. The labeling substance may further have, for example, a nucleic acid structure that does not contribute to interaction with the target substance (hereinafter also referred to as "other nucleic acid structures"), a quenching portion for controlling the fluorescence of the fluorescent portion, an artificial nucleic acid structure, a photoresponsive group, etc. Note that in this step, only one type of labeling substance may be mixed with the composition, or two or more types of labeling substances may be mixed in combination.

[0026] The interacting portion of the labeling substance has a nucleic acid structure that can interact nonspecifically with the target substance. Examples of interactions between the interacting portion and the target substance include hydrogen bonding between nucleic acids (hybridization), hydrogen bonding with protein amide groups or amino acids in biological substances such as enzymes and tumor markers, hydrogen bonding with acids, and interactions due to the shape of the formed three-dimensional structure.

[0027] The nucleic acid structure of the interaction portion is appropriately selected depending on the structure of the target object. When the target object is a nucleic acid, the nucleic acid structure preferably has a base sequence complementary to a portion of the base sequence of the target object. Furthermore, the complementary sequence of the nucleic acid structure preferably consists of a relatively short base sequence. A short nucleic acid structure (base sequence) facilitates interaction not only with the nucleic acid structure of one target object but also with the nucleic acid structures of multiple targets. Furthermore, when the target object contains multiple identical or similar base sequences, it facilitates interaction with each of these. However, if the number of bases in the complementary sequence within the nucleic acid structure is excessively small, it will affect many structures of the target object, resulting in a very complex fluorescent fingerprint. Therefore, the number of complementary bases in the nucleic acid structure of the interaction portion is preferably approximately 5 to 20, more preferably 7 to 10. A base number within this range facilitates appropriate interaction between the labeling substance and the target object. The labeling substance contains one or more such interaction portions (nucleic acid structures) within the molecule.

[0028] On the other hand, the structure and position of the fluorescence-emitting moiety contained in the labeling substance are not particularly limited, and for example, it may be located at a position separated from the above-mentioned interaction part via another nucleic acid structure. However, it is preferable that the fluorescence-emitting moiety is located close to the interaction part, and it is preferably located at the end of the interaction part or between bases of the nucleic acid structure of the interaction part. If the distance between the fluorescence-emitting moiety and the interaction part is short, the wavelength and intensity of the fluorescence emitted by the fluorescence-emitting moiety are likely to change depending on the state of interaction between the interaction part and the target substance, etc.

[0029] Furthermore, the number of fluorescent moieties contained in the labeling substance is not particularly limited and may be one or more. When the labeling substance has multiple fluorescent moieties, the fluorescent moieties can interact with each other to cause self-quenching when the labeling substance is not interacting with the target substance, or to increase the emission intensity when the labeling substance interacts with the target substance.

[0030] The type of dye contained in the fluorescent-emitting portion is not particularly limited, and examples include common fluorescent dyes (e.g., cyanine dyes, merocyanine dyes, acridine dyes, coumarin dyes, ethidium dyes, flavin dyes, condensed aromatic ring dyes, xanthene dyes, etc.).

[0031] Furthermore, the labeling substance may have an artificial nucleic acid structure between the bases of the nucleic acid structure. An artificial nucleic acid structure refers to a nucleic acid structure in which a non-natural portion has been introduced into a natural nucleic acid or which is synthesized solely with a non-natural portion. Artificial nucleic acid structures are completely artificially synthesized molecules that do not normally exist in nature. Therefore, they are characterized by being difficult to recognize and decompose by nucleases present in the air. Inserting such an artificial nucleic acid structure into the interaction portion can stabilize the labeling substance or increase the fluorescence intensity of the fluorescent portion. Examples of artificial nucleic acid structures include, but are not limited to, acyclic threoninol nucleic acid (aTNA), serinol nucleic acid (SNA), peptide nucleic acid (PNA), glyconucleic acid (GNA), locked nucleic acid (LNA), etc.

[0032] Furthermore, the labeling substance may contain an insulator between bases in the nucleic acid structure of the interacting portion to enhance the fluorescence intensity of the fluorescent-emitting portion. When the labeling substance contains an insulator, electron transfer to other fluorescent-emitting portions or the like is suppressed, making it easier to increase the fluorescence intensity of the fluorescent-emitting portion. The insulator is usually disposed adjacent to the fluorescent-emitting portion. The insulator can be, for example, a ring-shaped body having a non-planar structure, and examples thereof include structures derived from cyclobutane, cyclopentane, cyclohexane, cycloheptane, cyclopentene, cyclohexene, cycloheptene, derivatives thereof, and the like.

[0033] Furthermore, the labeling substance may have a photoresponsive group between the bases of the nucleic acid structure of the interaction portion, whose structure changes upon irradiation with light of a specific wavelength. When a photoresponsive group is present between the bases of the nucleic acid structure of the interaction portion, the interaction state between the labeling substance (interaction portion) and the target substance changes upon irradiation with specific light. As a result, the wavelength and intensity of the fluorescence emitted by the fluorescence-emitting portion can be more finely adjusted. Examples of the photoresponsive group include groups that change from a planar structure to a three-dimensional structure upon irradiation with specific light. Specifically, these include groups that change from a trans form to a cis form and groups that change from a merocyanine form to a spiropyran form. In particular, structures derived from azobenzene, spiropyran, stilbene, or derivatives thereof are preferred because their structure changes reversibly upon irradiation with light of a specific wavelength.

[0034] Here, the labeling substance may have a stem structure and a loop structure. When the labeling substance has a fluorescent-emitting moiety and a quenching moiety, the fluorescent-emitting moiety and the quenching moiety corresponding to the fluorescent-emitting moiety can be arranged in the stem structure so as to face each other. In other words, under normal conditions, the quenching moiety and the fluorescent-emitting moiety interact with each other to suppress the emission of the fluorescent-emitting moiety. On the other hand, when the labeling substance interacts with a target substance or the like, the stem structure opens, the quenching suppression by the quenching moiety is released, and the fluorescent-emitting moiety emits fluorescence. The quenching moiety is not particularly limited as long as it can interact with the fluorescent-emitting moiety and suppress its emission, and can be, for example, a structure derived from azobenzene or a derivative thereof.

[0035] When the labeling substance has a stem structure and a loop structure, the interacting part may be located within the loop structure, within the stem structure, or across both.

[0036] In cases where the target substance is an RNA vaccine, a labeling substance may be used in which one strand of the stem structure is a poly-U sequence, the other strand of the stem structure is a poly-A sequence, and the loop structure is an interaction site. In such a labeling substance, under normal conditions, the stem is formed by the poly-U sequence and the poly-A sequence. However, when mixed with the target substance, the end (e.g., the poly-A sequence) of the labeling substance (here, the complementary poly-U sequence) approaches, bringing other complementary base sequence portions into proximity and opening the stem. This facilitates hybridization, leading to improved efficiency. Alternatively, a labeling substance may be used in which one side of the stem structure is a fluorescent moiety and the other side is a quencher. Under normal conditions, the fluorescent moiety and quencher are in close proximity, suppressing the emission of the fluorescent moiety. However, when mixed with the target substance, the stem opens, separating the quencher, allowing the fluorescent moiety to emit fluorescence.

[0037] The labeling substance may be mixed directly with the composition, or the labeling substance encapsulated in particles such as gelatin nanoparticles or lipid nanoparticles may be mixed with the composition. When the labeling substance is encapsulated in gelatin nanoparticles or lipid nanoparticles, the nucleic acid structure within the labeling substance is less susceptible to the effects of atmospheric enzymes. Therefore, degradation of the labeling substance can be prevented, improving stability and enabling long-term storage. Since the nonspecifically acting labeling substance of this embodiment does not need to be adjusted for each measurement target, multiple types of labeling substances can be prepared and stored in advance, allowing for rapid response to new measurement targets.

[0038] When a labeled substance is encapsulated in gelatin nanoparticles or lipid nanoparticles, it usually cannot interact with the target substance as is. Therefore, the labeled substance must be released from the particles, which can be easily achieved by adding a surfactant or making the mixture acidic. Furthermore, when the target substance is an RNA vaccine, it is usually protected by lipid nanoparticles, etc., as mentioned above, so the encapsulated mRNA is similarly released and measured.

[0039] (2) Fluorescence fingerprint acquisition process In the fluorescence fingerprint acquisition step S12, a fluorescence fingerprint is acquired from the mixture. The fluorescence fingerprint is preferably three-dimensional data representing the wavelength of excitation light irradiated onto the mixture, the wavelength of fluorescence emitted by the mixture (the fluorescence-emitting portion) irradiated with the excitation light, and its intensity. The fluorescence fingerprint can be represented, for example, by graphs such as those shown in FIGS. 1A and 1B. Note that FIG. 1A is the fluorescence fingerprint of a specific resin, and FIG. 1B is the fluorescence fingerprint after the resin has been heated and denatured.

[0040] The method for acquiring the fluorescence fingerprint of the mixture is as follows. First, the mixture is irradiated with excitation light of a specific wavelength from an excitation light source. Then, the wavelength and intensity of the fluorescence emitted by the mixture when irradiated with the excitation light are measured. Next, the wavelength of the excitation light is shifted by a desired width (for example, 10 nm), and the wavelength and intensity of the fluorescence are measured in the same way. This is repeated. In this way, data on the fluorescence wavelength and fluorescence intensity corresponding to the excitation light in the desired wavelength range is obtained. Then, these are converted into three-dimensional data to acquire the fluorescence fingerprint.

[0041] The wavelength of the excitation light used to obtain the fluorescent fingerprint is selected appropriately depending on the type of fluorescent moiety in the labeling substance, the type of target object, etc. For example, if the target object is an RNA vaccine, visible light can be used. The light source used to irradiate the excitation light is not particularly limited, but can be a supercontinuum light source (a broadband pulsed light source that uses the nonlinear effect of optical fiber to emit strong, phase-aligned light over an extremely wide wavelength range, also known as an "SC light source") or an LED. These light sources can increase the amount of light, making it easier to obtain a good fluorescent fingerprint. Multiple light sources can also be combined.

[0042] On the other hand, the wavelength and intensity of the fluorescence emitted by the mixture can be measured using a spectrofluorometer, etc. The measurement may be performed using multiple spectrofluorometers.

[0043] Furthermore, the wavelength of the excitation light, the wavelength of the fluorescence, and the fluorescence intensity can be converted into three-dimensional data using a general personal computer or the like.

[0044] (3) Evaluation process In the evaluation step S13, the state of the composition is evaluated. In this specification, the state of the composition refers to the concentration of the target substance contained in the composition, the composition of the composition, the altered state or deteriorated state of a specific target substance, etc. For example, the evaluation step S13 includes the degree of deterioration or aggregation of the target substance, the degree of chemical or physical change of the target substance, the authenticity of the composition, and the suitability of use of the composition.

[0045] In particular, it has been difficult with conventional methods to identify the degree of deterioration of an object. Therefore, it is particularly preferable to evaluate the degree of deterioration of an object. Generally, when an object deteriorates, it is often difficult to separate the object from the deteriorated object, or the differences are too small to distinguish. In contrast, the method of the present embodiment can detect these small differences and perform a qualitative evaluation with high accuracy.

[0046] Here, the state of the composition may be evaluated by comparing the fluorescence fingerprint obtained in the fluorescence fingerprint acquisition step S12 described above with the fluorescence fingerprint of the object in an ideal state (for example, an object that is not deteriorated) to determine the degree of change in state. Alternatively, a fluorescence fingerprint of the object after a state change may be acquired, and the fluorescence fingerprint may be compared with this fluorescence fingerprint to determine the degree of change in state. These comparisons may be performed by the personal computer described above.

[0047] Furthermore, the fluorescence fingerprint obtained in the fluorescence fingerprint acquisition step S12 may be subjected to statistical analysis to reduce its dimensionality and parameterize it to clearly represent the characteristics of each state of the composition, thereby evaluating the composition. Examples of statistical analysis methods include multivariate analysis and data mining. Specific examples include data structure analysis, discriminant analysis, pattern classification, multidimensional data analysis, regression analysis, and machine learning.

[0048] The data structure analysis includes principal component analysis, factor analysis, correspondence analysis, and independent component analysis. The discriminant analysis includes linear discriminant analysis or nonlinear discriminant analysis. The linear discriminant analysis includes canonical discriminant analysis, and the nonlinear discriminant analysis includes decision trees.

[0049] Examples of the pattern classification include cluster analysis and multidimensional scaling. Examples of the regression analysis include linear regression and nonlinear regression. Here, examples of linear discriminant analysis include partial least squares (PLS) regression, simple regression analysis, multiple regression analysis, and principal component regression, and examples of nonlinear discriminant analysis include logistic regression and regression tree.

[0050] Examples of the machine learning include neural networks, self-organizing maps, ensemble learning, and genetic algorithms. The statistical analysis process may be performed using any analytical method that can most accurately analyze the state of the composition.

[0051] Furthermore, as in a modified example shown in the flow chart of FIG. 3, when evaluating the state of a composition, this process may include a step S100 of preparing a state estimation model in advance, and a step S130 of evaluating the state of the object based on the fluorescent fingerprint obtained in the fluorescent fingerprint acquisition step S12. Use of the state estimation model facilitates evaluation. A pre-prepared state estimation model can be used, and does not need to be created each time. Furthermore, in the flow chart of FIG. 3, the step S100 of preparing a state estimation model is described before the mixture preparation step S11, but the step S100 of preparing a state estimation model may be performed before or after the fluorescent fingerprint acquisition step S12.

[0052] The state estimation model can be created, for example, as follows. The creation flow of the state estimation model is shown in FIG. 4. First, a plurality of samples with different composition states are prepared (S101). At this time, if a detailed evaluation is desired, such as the deterioration rate of the target substance, a plurality of samples may be prepared in which the deterioration rate of the target substance in the composition has been adjusted. On the other hand, if it is difficult to finely adjust the state of the composition and a qualitative evaluation is to be performed, a sample with a large state change, a sample with a medium state change, a sample with a small state change, etc. may be prepared.

[0053] Then, similar to the mixture preparation process described above, a labeling substance is mixed with each of a plurality of samples with different composition states to prepare a standard mixture (S102). A standard fluorescence fingerprint is then obtained for each of these plurality of standard mixtures (S103). The method for obtaining the standard fluorescence fingerprint is the same as the fluorescence fingerprint obtaining process described above. After that, machine learning is performed on the obtained standard fluorescence fingerprint and the composition state to obtain a state estimation model (S104). Note that when creating a state estimation model, it is preferable to obtain many standard fluorescence fingerprints using not only one type (or one combination) of labeling substance, but also different types of labeling substances. For example, if there are five samples and ten types of labeling substances, 5 x 10 types of fluorescence fingerprints will be obtained by machine learning.

[0054] Known methods can be used for machine learning. For example, a multivariate analysis is performed on a specific sample using a standard fluorescent fingerprint as an explanatory variable and the degree of change in state as a target variable to determine a similarity index. The similarity index can be selected from cosine similarity, Pearson's correlation coefficient, deviation pattern similarity, Euclidean distance similarity, Morishita's similarity index, standard Euclidean distance similarity, Mahalanobis distance similarity, Manhattan distance similarity, Chebyshev distance similarity, Minkowski distance similarity, Jaccard coefficient similarity, Dice coefficient similarity, and Simpson coefficient similarity.

[0055] In the similarity index calculation step, under the assumption that there is a correlation between the degree of state change (the objective variable) and the similarity index of the fluorescence fingerprint of the object being measured, the similarity index of the fluorescence fingerprint is calculated for a certain sample of a certain degree of state change prepared in advance. This is performed for multiple samples with different degrees of state change, and an estimation model can be created by repeatedly executing and optimizing the process so that the error between the calculated similarity index and the degree of state change of the prepared sample is minimized.

[0056] Many analytical methods have been developed for two-dimensional data. The aforementioned fluorescence fingerprints, on the other hand, are three-dimensional data consisting of excitation light wavelength, fluorescence wavelength, and fluorescence intensity. Therefore, the three-dimensional data can be expanded into two dimensions for multivariate analysis. For example, the fluorescence fingerprint measurement results can be expanded into two-dimensional data for each labeled substance, consisting of wavelength conditions (combinations of excitation wavelength and fluorescence wavelength) and fluorescence intensity, and multivariate analysis can be performed. Furthermore, the two-dimensionally expanded fluorescence fingerprint information can be subjected to processes such as mean centering, normalization, autoscaling, second derivative, baseline correction, and smoothing. This allows for emphasizing the information contained in each data point and aligning the scale of data from different samples. Alternatively, multivariate analysis can be performed on the three-dimensional data as is.

[0057] Furthermore, if necessary, marker signals (e.g., combinations of excitation wavelength and fluorescence) important for estimation (i.e., those that change significantly depending on the state of the composition) can be detected from the obtained similarity index. In this case, marker signals may be detected by performing multivariate analysis such as principal component regression, cluster analysis, discriminant analysis, SIMCA, multiple regression analysis, PLS regression analysis, PLS discrimination, SVM regression, SVM discrimination, RF regression, and / or RF discrimination. Marker signals may also be detected based on one or more indices indicating the contribution rate to regression and discrimination, including regression coefficients, factor loadings, loadings, selectivity ratios, variable importance inprojection, variable importance, and out-of-bag error, obtained from multivariate analysis.

[0058] Then, the numerical value of the marker signal is estimated for a sample of a composition different from the sample for which the similarity index was calculated. The estimated data is then compared with the actual data, and optimization is performed repeatedly to minimize the error. Through these operations, a state estimation model capable of identifying the state of the composition from fluorescent fingerprint data, etc., may be obtained.

[0059] Then, by applying the fluorescence fingerprint obtained in the fluorescence fingerprint obtaining step to the state estimation model, the state of the composition can be evaluated.

[0060] (effect) In this embodiment, a labeling substance that interacts nonspecifically with the target substance is used, and a fluorescent fingerprint is further obtained. According to the method of this embodiment, the state of a composition can be grasped in a simple manner without separating the target substance or the like in the composition. Furthermore, even if there are multiple targets, this method does not require preparing a labeling substance for each target substance, reducing the cost and time required for preparing the labeling substance. Furthermore, even if the target substance changes to an unexpected structure, there is a high possibility that it can be detected using the labeling substance. Therefore, for example, mRNA vaccines and the like can be evaluated and identified using a very simple process.

[0061] 2. Second embodiment The flow of the composition evaluation method of this embodiment is shown in Figure 5. The composition evaluation method of this embodiment includes the steps of: preparing a mixture by mixing a composition containing one or more target substances and a labeling substance having an interaction moiety and a fluorescence-emitting moiety in step S21 (hereinafter also referred to as the "mixture preparation step"); irradiating the mixture with predetermined excitation light and measuring the fluorescence in step S22 (hereinafter also referred to as the "fluorescence measurement step"); and evaluating the state of the composition from the measurement results in step S23 (hereinafter also referred to as the "evaluation step"). In this embodiment, as in the first embodiment, a substance that acts nonspecifically on the target substance is used as the labeling substance.

[0062] Furthermore, the wavelength of the excitation light irradiated in the fluorescence measurement step of this embodiment is determined in advance by performing a step (S200) of determining a predetermined wavelength from a state estimation model. That is, in this embodiment, a wavelength of excitation light suitable for detecting changes in the state is selected based on a prior analysis of how the emission wavelength and emission intensity of the mixture change when the state of the composition changes. Therefore, in the fluorescence measurement step, the fluorescence wavelength and fluorescence intensity generated by excitation light of a predetermined wavelength are measured and compared with reference values, etc., to easily evaluate the state of the composition. In the flow chart shown in FIG. 5, the step S200 of determining the wavelength of the excitation light is described before the mixture preparation step S21. However, once the step S200 of determining the wavelength of the excitation light has been performed once, it is not necessary to perform the wavelength determination step S200 each time. Therefore, if the wavelength of the excitation light has already been determined, only the mixture preparation step S21, the fluorescence measurement step S22, and the evaluation step S23 are performed. The mixture preparation step S21 of this embodiment is similar to the mixture preparation step S11 of the first embodiment described above, and therefore the fluorescence measurement step and evaluation step will be described below.

[0063] (1) Fluorescence measurement process In the fluorescence measurement step S22, the mixture obtained in the mixture preparation step S21 is irradiated with a predetermined excitation light, and the fluorescence emitted by the mixture is measured. The excitation light of a predetermined wavelength irradiated in this step is the excitation light identified by creating a state estimation model, as described above. In this step, only one type of excitation light may be irradiated and the fluorescence emitted by the mixture may be measured. However, from the viewpoint of enabling more accurate evaluation, it is preferable to irradiate the mixture with excitation light of multiple wavelengths in sequence and measure the fluorescence emitted by the mixture for each wavelength.

[0064] The light source of the excitation light in this step is appropriately selected depending on the wavelength of the excitation light, and can be an SC light source, an LED, etc. The fluorescence measuring device can be a spectrofluorometer, a one-dimensional line sensor, etc.

[0065] The method for creating a state estimation model to determine the wavelength of the excitation light to be irradiated onto the mixture in this step is substantially the same as the method described in the evaluation step S13 of the first embodiment. Figure 6 shows a flow for determining the wavelength of the excitation light. First, multiple samples with different composition states are prepared (S201). Then, similar to the mixture preparation step S21 described above, a labeling substance is mixed with each of the multiple samples to prepare a standard mixture (S202). Then, a standard fluorescence fingerprint is obtained for each of the multiple standard mixtures (S203). Then, machine learning is performed on the obtained standard fluorescence fingerprint and the composition state (S204). Then, one or more marker signals (e.g., combinations of excitation wavelength and fluorescence wavelength) that are important for estimation (i.e., that change significantly depending on the state of the composition) are detected from the similarity index calculated based on the standard fluorescence fingerprint and the composition state (S205). Then, one or more excitation light wavelengths from the marker signals are selected, and these are used as the excitation light wavelengths in the fluorescence measurement step S22.

[0066] (2) Evaluation process In the evaluation step S23, the state of the composition is evaluated based on the data obtained in the fluorescence measurement step S22. For example, information is obtained in advance from the state estimation model, such as information that if fluorescence at wavelength λ2 is detected at excitation wavelength λ1, the degree of degradation of the target substance in the composition is high, or information that if the intensity of fluorescence wavelength λ4 is high at excitation wavelength λ3, the composition is highly deteriorated. The state of the composition can then be evaluated by comparing this information with the data on the wavelength and intensity of the fluorescence measured in the fluorescence measurement step. Alternatively, the data obtained in the fluorescence measurement step S22 may be applied to the state estimation model created when determining the predetermined wavelengths to evaluate the state of the composition.

[0067] (3) Effects In this embodiment, a labeling substance that interacts nonspecifically with the target substance is used. Furthermore, fluorescence measurement is performed by irradiating the substance with an excitation wavelength that is determined in advance based on a state estimation model. Therefore, in this embodiment, it is not necessary to measure the entire wavelength range during measurement, and the state of the target composition can be easily evaluated with a short measurement time.

[0068] Furthermore, this method does not require the preparation of a labeling substance for each target object, reducing the cost and time required for preparing the labeling substance. Furthermore, even if the target object undergoes an unexpected structural change, it is highly likely that the change can be detected using the labeling substance. Therefore, for example, mRNA vaccines can be evaluated and identified through a very simple process.

[0069] 3.Other In any of the above-described embodiments, the method for mixing the composition and the labeling substance in the mixture preparation step or the like is not particularly limited. For example, a plate having a plurality of recesses, each containing a different labeling substance, may be prepared, and a predetermined amount of the composition may be injected into the recesses by an inkjet method. The plate may then be used to perform the fluorescence fingerprint acquisition step and the fluorescence measurement step. This method allows for the easy preparation of a plurality of mixtures and a standard mixture.

[0070] This application claims priority from Japanese Patent Application No. 2021-126668, filed August 2, 2021. The contents of the specification and drawings of that application are incorporated herein by reference in their entirety. [Industrial Applicability]

[0071] The composition evaluation method of the present invention makes it possible to grasp the state of a composition in a simple manner without separating the object, and is therefore useful for measuring the degree of deterioration of various medicines and foods, and for testing in various medical fields.

Claims

1. a step of mixing a composition containing one or more target substances and a labeling substance having an interacting moiety and a fluorescent moiety to prepare a mixture; obtaining a fluorescence fingerprint of the mixture; assessing the condition of the composition based on the fluorescent fingerprint; Including, the interacting part has a nucleic acid structure capable of interacting with each of a plurality of types of the target substance, and / or a nucleic acid structure capable of interacting with a plurality of positions of each of one or more types of the target substance, In the step of evaluating the condition, a degree of deterioration of the composition is evaluated. Methods for evaluating compositions.

2. A method for detecting a target substance comprising: mixing a composition containing one or more target substances and a labeling substance having an interacting portion and a fluorescent portion to prepare a mixture; obtaining a fluorescence fingerprint of the mixture; assessing the condition of the composition based on the fluorescent fingerprint; Including, the interacting part has a nucleic acid structure capable of interacting with each of a plurality of types of the target substance, and / or a nucleic acid structure capable of interacting with a plurality of positions of each of one or more types of the target substance, the interacting part has a photoresponsive group between bases of the nucleic acid structure, the structure of which changes when irradiated with light of a specific wavelength; Methods for evaluating compositions.

3. using a plurality of types of labeling substances each having a different interacting portion; A method for evaluating the composition according to claim 1 or 2.

4. In the step of evaluating the state, the state of the object is evaluated based on a state estimation model; The state estimation model is preparing a plurality of samples each having a different state of the composition; preparing a plurality of standard mixtures by mixing the labeling substance with each of the plurality of samples; obtaining a standard fluorescence fingerprint for each of the plurality of standard mixtures; The standard fluorescence fingerprints are created by machine learning the state of the composition in each sample and each standard fluorescence fingerprint. A method for evaluating the composition according to any one of claims 1 to 3.

5. The interaction portion includes an artificial nucleic acid structure. A method for evaluating the composition according to any one of claims 1 to 4.

6. the labeling substance has the fluorescent moiety between the bases of the interacting portion; A method for evaluating the composition according to any one of claims 1 to 5.

7. the labeling substance has a loop structure and a stem structure, the fluorescent-emitting portion and the quenching portion corresponding to the fluorescent-emitting portion are disposed opposite to each other within the stem structure; A method for evaluating the composition according to claim 6.

8. The labeling substance is encapsulated in a particle. A method for evaluating the composition according to any one of claims 1 to 7.

9. a step of mixing a composition containing one or more target substances and a labeling substance having an interacting moiety and a fluorescent moiety to prepare a mixture; irradiating the mixture with excitation light of a predetermined wavelength and measuring the fluorescence emitted from the mixture; evaluating the state of the object from the results of the measurement; Including, In the step of evaluating the state of the object, a degree of deterioration of the composition is evaluated; the interactive portion of the labeling substance has a nucleic acid structure capable of interacting with each of the plurality of types of target substances, and / or has a nucleic acid structure capable of interacting with each of the plurality of positions of the target substances, The wavelength of the excitation light is preparing a plurality of samples each having a different state of the composition; preparing a plurality of standard mixtures by mixing the labeling substance with each of the plurality of samples; obtaining a standard fluorescence fingerprint for each of the plurality of standard mixtures; The wavelengths are identified from a state estimation model created by machine learning the state of the composition in each sample and each of the standard fluorescence fingerprints. Methods for evaluating compositions.

10. A method for detecting a target substance comprising: mixing a composition containing one or more target substances and a labeling substance having an interacting portion and a fluorescent portion to prepare a mixture; irradiating the mixture with excitation light of a predetermined wavelength and measuring the fluorescence emitted from the mixture; evaluating the state of the object from the results of the measurement; Including, the interactive part of the labeling substance has a nucleic acid structure capable of interacting with each of the plurality of types of target substances, and / or has a nucleic acid structure capable of interacting with each of the target substances at a plurality of positions, and has a photoresponsive group between bases of the nucleic acid structure, the structure of which changes upon irradiation with light of a specific wavelength; The wavelength of the excitation light is preparing a plurality of samples each having a different state of the composition; preparing a plurality of standard mixtures by mixing the labeling substance with each of the plurality of samples; obtaining a standard fluorescence fingerprint for each of the plurality of standard mixtures; The wavelengths are identified from a state estimation model created by machine learning the state of the composition in each sample and each of the standard fluorescence fingerprints. Methods for evaluating compositions.

Images