JP7268530B2 - Mass spectrometry data processing method, mass spectrometry data processing system, and mass spectrometry data processing program - Google Patents

Description

The present invention relates to a mass spectrometry data processing method, a mass spectrometry data processing system, and a mass spectrometry data processing program.

Matrix Assisted Laser Desorption/Ionization (MALDI) is well known as one of ionization methods for mass spectrometers. In the MALDI method, in order to analyze samples that do not easily absorb laser light or that are easily damaged by laser light, such as proteins, substances that easily absorb and ionize laser light are premixed with the sample as a matrix. In this method, a sample is ionized by irradiating it with a laser beam. In particular, a mass spectrometer using a MALDI ion source (hereafter referred to as MALDI-MS) is capable of analyzing high-molecular-weight compounds with little cleavage, and is also suitable for microanalysis. Since then, it has been widely used in fields such as life science.

In recent years, attempts have been made to discriminate unknown samples by applying machine learning to mass spectra obtained by MALDI-MS (see, for example, Patent Document 1). Machine learning is one of the useful methods for discovering regularity from a large amount of diverse data and using it for prediction, discrimination, or regression of data. and unsupervised learning. For example, when trying to discriminate the type of microorganism (e.g., species, subspecies, strain, or type) based on the results of analyzing microorganisms by MALDI-MS, a large amount of mass spectrometry data for various microorganisms is prepared in advance. Collected data are used as learning data (also referred to as teacher data or training data) to perform supervised learning to construct a discriminant model for discriminating the types of unknown microorganisms.

JP 2018-155522 A Japanese Patent Application Laid-Open No. 2010-205460

However, in order to construct a highly accurate discriminant model, it is necessary to collect a large amount of learning data. For that purpose, it is necessary to perform mass spectrometry many times, and there is a problem that much labor and cost are required.

The present invention has been made in view of the above points, and its object is to obtain a large amount of learning data necessary for constructing a highly accurate discriminant model with a small number of mass spectrometry. An object of the present invention is to provide an analysis data processing method, a mass spectrometry data processing system, and a mass spectrometry data processing program.

The mass spectrometry data processing method according to the present invention, which has been made to solve the above problems,
A relationship between m/z and intensity of ions generated from the known sample in each of the plurality of times of laser light irradiation in a mass spectrometer that ionizes the sample by laser ionization. Acquire multiple profile data that are spectra showing
sorting the plurality of profile data into a plurality of groups such that each group contains one or more profile data;
For each of the plurality of groups, generating a peak list that describes the m/z of the peaks derived from the known sample and the intensities of the peaks based on the one or more profile data included in the group;
A discriminant model for discriminating an unknown sample is generated by using the peak list and information on the type of the known sample as learning data.

The mass spectrometry data processing system according to the present invention, which was made to solve the above problems,
m/z of ions generated from the known sample at each of the plurality of laser light irradiations, obtained by irradiating the known sample with the laser light multiple times in a mass spectrometer that ionizes the sample by laser ionization; and a profile data acquisition unit that acquires a plurality of profile data, which are spectra showing a relationship with intensity;
a grouping unit that sorts the plurality of profile data into a plurality of groups such that each group includes one or more profile data;
A peak list for generating a peak list describing the m/z of the peak derived from the known sample and the intensity of the peak for each of the plurality of groups based on the one or more profile data included in the group. a generator;
a discriminant model generation unit that generates a discriminant model for discriminating an unknown sample using the peak list and information about the type of the known sample as learning data;
is provided.

A mass spectrometry data processing program according to the present invention, which has been made to solve the above problems, causes a computer to function as each part of the mass spectrometry data processing system.

In the mass spectrometry data processing method, the mass spectrometry data processing system, and the mass spectrometry data processing program according to the present invention, profile data obtained by irradiating one sample with laser light many times is divided into a plurality of groups. , to generate one peak list for each group. This makes it possible to increase the number of peak lists obtained by mass spectrometry for one sample. As a result, it is possible to obtain a large amount of learning data necessary for constructing a highly accurate discriminant model with a small number of mass spectrometry runs.

1 is a block diagram showing the main configuration of a mass spectrometry data processing system according to one embodiment of the present invention; FIG. 4 is a flowchart showing a procedure for processing mass spectrometry data in the same embodiment;

EMBODIMENT OF THE INVENTION Hereinafter, it demonstrates, referring drawings for the form for implementing this invention. FIG. 1 is a block diagram showing the essential configuration of a mass spectrometry data processing system 10 according to one embodiment of the present invention.

This system 10 processes mass spectrometry data obtained by analyzing a sample by MALDI-MS (not shown), and includes a learning data generation unit 20, a discrimination model generation unit 30, a discrimination unit 40, and data storage. 50, an input unit 60 including a pointing device such as a mouse and a keyboard, and a display unit 70 including a display device such as a liquid crystal display.

The learning data generation unit 20 generates learning data for use in machine learning by performing predetermined processing on mass spectrometry data obtained by analyzing a known sample (for example, a microorganism whose strain is known) by MALDI-MS. is generated. The learning data generation unit 20 includes a profile data acquisition unit 21 , a grouping unit 22 and a peak list generation unit 23 .

The discriminant model generation unit 30 uses a plurality of learning data generated by the learning data generation unit 20 to generate a discriminant model for discriminating an unknown sample (for example, a microorganism whose strain is unknown).

The discrimination unit 40 discriminates the type of the unknown sample (for example, the strain to which the microorganism belongs) by applying mass spectrometry data obtained by analyzing the unknown sample by MALDI-MS to the discrimination model. . The determination unit 40 includes an unknown sample data acquisition unit 41 and a determination execution unit 42 .

The entity of the learning data generation unit 20, the discriminant model generation unit 30, and the discrimination unit 40 is a computer (personal computer or a computer with higher performance than that), and dedicated data processing software pre-installed in the computer is used. The functions of the respective units are realized by operating them on a computer. The data storage unit 50 may be a storage device built into the computer or directly connected to the computer, or may exist on another computer system accessible from the computer via the Internet or the like, for example. , that is, a storage device or the like in cloud computing may be used.

In addition, the system 10 according to the present embodiment can also share the functions of the learning data generating unit 20, the discriminant model generating unit 30, and the discriminating unit 40 among a plurality of computers. Specifically, for example, it is conceivable to assign the functions of the learning data generator 20 and the discriminant model generator 30 to one computer, and assign the function of the discriminator 40 to another computer.

Next, features of processing in the system 10 according to this embodiment will be described.

Generally, in MALDI-MS, the process of generating ions by laser light irradiation → separating and detecting the generated ions is repeatedly executed many times (for example, 120 times) for one sample, and many profiles are obtained. Data is generated (see Patent Document 2, etc.). Profile data is a data format equivalent to raw data of a mass spectrometer. The waveform of the detection signal continuously sent from the ion detector provided in the mass spectrometer is plotted on the horizontal axis with time. (or m/z), and the vertical axis is the ion intensity.

In the conventional data processing method, all the profile data obtained by irradiating a single sample with multiple laser beams as described above are integrated, and for the convenience of subsequent data processing, after the integration, Peaks included in the waveform of the profile data (integrated profile data) are detected (that is, peak detection processing is performed), and the m/z value representing the centroid position (or center position) of each detected peak and the peak were converted to a list (peak list) showing the area values of . That is, in the conventional data processing method, one peak list is generated as a result of one-time mass spectrometric analysis for one sample.

On the other hand, the mass spectrometry data processing method according to the present embodiment divides the profile data obtained by irradiating one sample with laser light many times as described above into a plurality of groups, generate a single peak list. That is, multiple peak lists are generated as a result of one-time mass spectrometric analysis for one sample. This makes it possible to obtain more learning data without increasing the number of mass spectrometry runs.

The details of such processing will be described below with reference to the flowchart of FIG. Here, mass spectrometry by MALDI-MS is performed on a plurality of known samples (for example, microorganisms whose strains are known) in advance, and N (N is 2 (integers above) are stored in the data storage unit 50 in association with information on the type of known sample (for example, information on strains of known microorganisms). Hereinafter, the information on the type of known sample will be referred to as a "correct label".

First, when the user performs a predetermined operation on the input unit 60 to specify the mass spectrometry results of the plurality of known samples stored in the data storage unit 50 and instructs the generation of learning data based on these, learning The data generation unit 20 generates learning data. Specifically, first, the profile data acquisition unit 21 of the learning data generation unit 20 obtains the mass spectrometry result regarding one known sample among the mass spectrometry results of a plurality of known samples designated by the user, that is, the N profile data are obtained from the data storage unit 50 (step S11).

Next, the grouping unit 22 allocates the N pieces of profile data to predetermined M pieces (M is an integer equal to or less than N) according to a predetermined standard (for example, in order of generation of the profile data). (Step S12). At this time, each of the M groups includes at least one piece of profile data. Also, the number of pieces of profile data assigned to each group should be as uniform as possible. Note that the number M of groups may be a value stored in the system 10 in advance, or may be set freely by the user. Alternatively, it may be automatically determined by the system 10 based on the number N of profile data or the required discrimination accuracy .

In ionization of a sample by MALDI, if the same position on the sample is repeatedly irradiated with laser light, ions will gradually cease to be generated. The sample or the laser beam is moved so as to irradiate the light, and profile data is acquired for each different position (measurement point). At this time, the amount of ions generated from each measurement point varies depending on the density of the sample components in the measurement area. Therefore, in step S12, it is desirable to randomly allocate the N pieces of profile data to the M groups. As a result, appropriate learning data can be generated without being affected by the density of the sample components within the measurement area.

Further, in step S12, part or all of the N pieces of profile data may be redundantly allocated to a plurality of groups. In this way, even if the number N of profile data is small or the number M of groups is large, the number of profile data allocated to each group can be increased, thereby preventing a decrease in S/N. can be done.

Subsequently, the peak list generator 23 generates a peak list for each of the M groups generated in step S12 (step S13). Specifically, the peak list generator 23 checks the number of pieces of profile data included in each group, and for a group containing multiple pieces of profile data, integrates the pieces of profile data to generate integrated profile data. . After performing noise removal processing (background removal processing and smoothing processing) on the integrated profile data, peak detection is performed using a predetermined peak detection algorithm. Then, the centroid position or center position of the detected peak and the area value of the peak are obtained, and the m / z of the centroid position (or center position) of each peak and the area value (corresponding to the intensity) of the peak are described. Generate a peak list. On the other hand, for a group containing only one profile data, noise removal processing (background removal processing and smoothing processing) and peak detection processing are performed on the one profile data without performing the integration processing. Go to Generate a peak list. The obtained M peak lists (that is, the same number as the number of croup) are stored in the data storage unit 50 in association with the correct label.

After that, the processes of steps S11 to S13 are performed for all of the plurality of known samples designated by the user, and M peak lists are generated for each of the known samples. Here, for the sake of simplification of explanation, it is assumed that N profile data are obtained for all known samples, the profile data are divided into M groups, and a peak list is generated for each group. , the number N of profile data, and the number M of groups (and peak lists) may differ for each sample.

Subsequently, when the user operates the input unit 60 to instruct generation of a discriminant model using the peak list generated for each of the known samples as learning data, the discriminant model generation unit 30 generates a discriminant model. (Step S14). Specifically, the discriminant model generation unit 30 reads out each of the M peak lists generated for each of the known samples and the correct label associated with each of the peak lists from the data storage unit 50, and learns them. As data, a discriminant model is generated by a predetermined machine learning method. The generated discriminant model is stored in the data storage unit 50 . Note that the peak list in this embodiment is multidimensional data in which the m/z of each peak is one dimension, and the discriminant model is, for example, a discriminant analysis function representing the relationship between multidimensional input and output. .

The machine learning method used to generate the discriminant model in step S14 is not particularly limited as long as it performs supervised learning. do it. It is preferable to appropriately select the method to be used depending on the type and properties of the data to be analyzed.

After that, an unknown sample (for example, a microorganism whose strain is unknown) to be determined is analyzed by MALDI-MS, and the obtained peak list is stored in the data storage unit 50, and the user inputs through the input unit 60 to instruct the discrimination of the unknown sample by the discrimination model. In addition, the peak list of the unknown sample is obtained by integrating all a plurality of profile data obtained by analyzing the unknown sample by MALDI-MS, and performing background removal processing, smoothing processing, and peak detection on the integrated profile data. It is generated in advance by performing processing. In the determination unit 40 that has received the instruction from the user, the unknown sample data acquisition unit 41 reads the peak list of the unknown sample from the data storage unit 50 (step S15), and the determination execution unit 42 applies the unknown sample to the determination model. From the output values obtained by inputting the sample peak list, the type of the unknown sample (for example, the strain to which the unknown microorganism belongs) is discriminated (step S16).

The determination result by the determination unit 40 is stored in the data storage unit 50 and displayed on the screen of the display unit 70 to be presented to the user (step S17).

Note that the mass spectrometry data discrimination system and mass spectrometry data processing method according to the present embodiment are limited to generation of discrimination models for discrimination of microorganisms (discrimination of species, subspecies, strains, types, etc. to which unknown microorganisms belong). First, discrimination of various samples, for example, discrimination of oil type, discrimination of disease (biological samples derived from people suffering from a predetermined disease such as cancer and those derived from people not suffering from the disease It can be applied to generation of a discriminant model for discrimination from a biological sample. In addition, in the mass spectrometry data discrimination system and mass spectrometry data processing method according to the present embodiment, the profile data used to generate learning data and the peak list of unknown samples to be discriminated are obtained by analysis by MALDI-MS. Without limitation, it may be obtained by a mass spectrometer that ionizes a sample by other laser ionization methods, such as Surface Assisted Laser Desorption/Ionization.

The effect of the present invention was verified by the ability to discriminate between two types of microorganisms (group A and group B). Here, group A is Escherichia coli, and group B is microorganisms of the genus Achromobacter (Achromobacter. sp).

First, samples of group A and samples of group B were each measured four times by MALDI-MS. At this time, 120 pieces of profile data were obtained by irradiating the sample with laser 120 times for each measurement. Then, as an example, a peak list was generated by processing the profile data by the method of the present invention, and a discriminant model was generated using the peak list. As a comparative example, a peak list was generated by processing the profile data by a conventional method, and a discriminant model was generated using the peak list.

Specifically, in the example, 120 pieces of profile data obtained by one measurement were randomly divided into four groups when generating the discriminant model. Then, 30 pieces of profile data included in each group were integrated, noise removal processing and peak detection processing were performed on the obtained integrated profile data, and a single peak list was generated. Using the 32 peak lists (2 groups x 4 measurements x 4 groups) obtained as learning data, a discriminant model for discriminating between group A and group B was generated.

On the other hand, in the comparative example, when generating the discriminant model, all 120 pieces of profile data obtained in one measurement were integrated, and noise removal processing and peak detection processing were performed on the obtained integrated profile data. A single peak list was generated. Using the 8 peak lists (2 groups x 4 measurements) obtained as a learning model, a discriminant model for discriminating between group A and group B was generated.

In both Examples and Comparative Examples, statistical analysis software eMSTAT Solution (registered trademark) was used to generate a discriminant model, and SVM (support vector machine) was used as a machine learning algorithm (hereinafter the same). .

When the discriminant performance of the discriminant model of the example and the discriminant model of the comparative example was verified, both methods were 100% correct regarding the output result (whether the data is group A or group B) given test data. However, the error (estimation error) due to cross-validation was 13% in the model of the comparative example, whereas it was 0% in the model of the example. The leave-one-out method was used for the cross-validation (the same applies to Examples 2 and 3 described later). That is, one piece of data was extracted as test data from the learning data of each group, and machine learning was performed using the remaining data. This was repeated until all data became test data once, and the results were averaged to obtain an estimation error. As a result, it was confirmed that according to the present invention, a discriminant model with higher accuracy than the conventional one can be obtained without increasing the number of measurements.

In addition, as a further example (Example 2), 120 profile data, which is data for one measurement out of four measurements for each of the samples of Group A and Group B, are divided into 120 groups. divided into Then, noise removal processing and peak detection processing were performed on one piece of profile data included in each group to generate a peak list. A discriminant model for discriminating between group A and group B was generated using the 240 (2 groups x 1 measurement x 120 groups) peak list obtained as learning data. The reason why profile data for only one measurement for each group was used for the discriminant model here is to prevent processing from taking too much time due to an excessive amount of data.

In addition, as a further example (Example 3), 120 profile data obtained in each measurement were randomly divided into two groups for each of the four measurements for the samples in Group A and the samples in Group B. split. Then, 60 pieces of profile data included in each group were integrated, and a peak list was generated by subjecting the obtained integrated profile data to noise removal processing and peak detection processing. Using the 16 peak lists (2 groups x 4 measurements x 2 groups) obtained as learning data, a discriminant model for discriminating between group A and group B was generated.

When the discriminant performance of the discriminant models obtained in Examples 2 and 3 was verified, it was confirmed that a model with an estimation error of 0% could be generated and that the test data could be correct 100% in both cases.

[Various aspects]
It will be appreciated by those skilled in the art that the exemplary embodiments described above are specific examples of the following aspects.

(Section 1) A mass spectrometry data processing method according to one aspect includes:
A relationship between m/z and intensity of ions generated from the known sample in each of the plurality of times of laser light irradiation in a mass spectrometer that ionizes the sample by laser ionization. Acquire multiple profile data that are spectra showing
sorting the plurality of profile data into a plurality of groups such that each group contains one or more profile data;
For each of the plurality of groups, generating a peak list that describes the m/z of the peaks derived from the known sample and the intensities of the peaks based on the one or more profile data included in the group;
A discriminant model for discriminating an unknown sample is generated by using the peak list and information on the type of the known sample as learning data.

(Section 2) The mass spectrometry data processing method according to Section 1,
The plurality of profile data may be randomly distributed to the plurality of groups.

(Section 3) The mass spectrometry data processing method according to Section 1 or 2,
When allocating the plurality of profile data to the plurality of groups, at least one of the plurality of profile data may be redundantly allocated to two or more of the plurality of groups.

(Section 4) The mass spectrometry data processing method according to any one of Sections 1 to 3,
Furthermore, the unknown sample may be discriminated by applying a peak list generated based on profile data obtained by mass spectrometric analysis of the unknown sample to the discriminant model.

(Section 5) A mass spectrometry data processing system according to one aspect,
m/z of ions generated from the known sample at each of the plurality of laser light irradiations, obtained by irradiating the known sample with the laser light multiple times in a mass spectrometer that ionizes the sample by laser ionization; and a profile data acquisition unit that acquires a plurality of profile data, which are spectra showing a relationship with intensity;
a grouping unit that sorts the plurality of profile data into a plurality of groups such that each group includes one or more profile data;
A peak list for generating a peak list describing the m/z of the peak derived from the known sample and the intensity of the peak for each of the plurality of groups based on the one or more profile data included in the group. a generator;
a discriminant model generation unit that generates a discriminant model for discriminating an unknown sample using the peak list and information about the type of the known sample as learning data;
is provided.

(Section 6) The mass spectrometry data processing system according to Section 5,
The grouping unit may randomly sort the plurality of profile data into the plurality of groups.

(Section 7) The mass spectrometry data processing system according to Section 5 or 6,
The grouping unit may distribute at least one of the plurality of profile data to two or more of the plurality of groups in duplicate.

(Section 8) The mass spectrometry data processing system according to any one of Sections 5 to 7,
a discrimination unit that discriminates the unknown sample by applying a peak list generated based on profile data obtained by mass spectrometry of the unknown sample to the discriminant model;
may be further provided.

(Item 9) A mass spectrometry data processing program according to one aspect causes a computer to function as each part of the mass spectrometry data processing system according to any one of the 5th to 8th items.

According to the mass spectrometry data processing method according to item 1, the mass spectrometry data processing system according to item 5, or the mass spectrometry data processing program according to item 9, in order to construct a highly accurate discrimination model It is possible to obtain a large amount of required learning data with a small number of mass spectrometry runs.

In addition, according to the mass spectrometry data processing method described in item 2 or the mass spectrometry data processing system described in item 6, appropriate learning data is obtained without being affected by the density of the sample components in the measurement area on the sample. can be generated.

Further, according to the mass spectrometry data processing method described in paragraph 3 or the mass spectrometry data processing system described in paragraph 7, even if the number of profile data is small or the number of groups is large, Since the number of profile data to be stored can be increased, a decrease in S/N can be prevented.

10 Mass spectrometry data processing system 20 Learning data generation unit 21 Profile data acquisition unit 22 Grouping unit 23 Peak list generation unit 30 Discriminant model generation unit 40 Discrimination unit 41 Unknown sample data acquisition unit 42 Discrimination Execution unit 50 Data storage unit 60 Input unit 70 Display unit

Claims (9)

In a mass spectrometer that ionizes a sample by laser ionization, one known sample is irradiated with laser light a plurality of times, and the m/z of ions generated from the known sample at each of the plurality of laser light irradiations a profile data acquisition step of acquiring a plurality of profile data, which are spectra showing a relationship with intensity;
a grouping step of allocating the plurality of profile data into a plurality of groups such that each group includes one or more profile data;
For each of the plurality of groups, generate a peak list describing the m/z of the peak derived from the one known sample and the intensity of the peak based on the one or more profile data included in the group. a peak list generation step ;
a learning data generating step of generating learning data by associating information about the type of the one known sample with each of the plurality of peak lists related to the one known sample generated in the peak list generating step;
Discriminating an unknown sample using a plurality of learning data obtained by executing the profile data acquiring step, the grouping step, the peak list generating step, and the learning data generating step for each of the plurality of known samples A discriminant model generation step for generating a discriminant model for
A mass spectrometry data processing method comprising : The mass spectrometry data processing method according to claim 1, wherein the plurality of profile data are randomly assigned to the plurality of groups. 3. The method according to claim 1, wherein when the plurality of profile data are assigned to the plurality of groups, at least one of the plurality of profile data is redundantly assigned to two or more of the plurality of groups. mass spectrometry data processing method. Further, the unknown sample is discriminated by applying a peak list generated based on profile data obtained by mass spectrometry of the unknown sample to the discrimination model. Mass spectrometry data processing method. In a mass spectrometer that ionizes a sample by laser ionization, one known sample is irradiated with laser light a plurality of times, and the number of ions generated from the one known sample at each of the plurality of laser light irradiations. a profile data acquisition unit that acquires a plurality of profile data, which are spectra showing the relationship between m/z and intensity;
a grouping unit that sorts the plurality of profile data into a plurality of groups such that each group includes one or more profile data;
For each of the plurality of groups, generate a peak list describing the m/z of the peak derived from the one known sample and the intensity of the peak based on the one or more profile data included in the group. a peak list generator;
a learning data generation unit that generates learning data by associating information regarding the type of the one known sample with each of the plurality of peak lists related to the one known sample generated by the peak list generation unit;
An unknown sample using a plurality of learning data obtained by executing processing by the profile data acquisition unit, the grouping unit, the peak list generation unit, and the learning data generation unit for each of a plurality of known samples a discriminant model generation unit that generates a discriminant model for discriminating
A mass spectrometry data processing system comprising: The mass spectrometry data processing system according to claim 5, wherein the grouping unit randomly distributes the plurality of profile data to the plurality of groups. The mass spectrometry data processing system according to claim 5 or 6, wherein the grouping unit redundantly sorts at least one of the plurality of profile data into two or more of the plurality of groups. a discrimination unit that discriminates the unknown sample by applying a peak list generated based on profile data obtained by mass spectrometry of the unknown sample to the discriminant model;
The mass spectrometry data processing system according to any one of claims 5 to 7, further comprising: A mass spectrometry data processing program that causes a computer to function as each part of the mass spectrometry data processing system according to any one of claims 5 to 8.

Images