JP2009008582A - Chromatogram data processor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To accurately detect a peak of a target compound appearing in a mass chromatogram where there is a large baseline drift or many noises overlap. <P>SOLUTION: When the presence of a baseline drift is determined based on a signal duration width on the chromatogram (S2 and S3), signal intensity variation accompanying a lapse of time course is calculated, and a baseline is estimated by filtering with a median filter and a low pass filter of the bandwidth according to this variation (S6). A noise peak value is derived by subtracting the baseline from the chromatogram; normalization is performed using the fact that noise amplitude is stochastically proportional to the square root of the baseline; the standard deviation of noise distribution is determined based on the signal distribution of the normalized value; and the threshold for separating the noise from the peak based on the standard deviation is set (S6). Thus, detecting the peak using a statistical method allows accurate peak detection even if there is drift or noise of the baseline. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a data processing apparatus that processes chromatogram data obtained by chromatographic analysis, and in particular, chromatographic mass such as a gas chromatograph mass spectrometer (GC / MS) and a liquid chromatograph mass spectrometer (LC / MS). The present invention relates to a data processing apparatus suitable for detecting a compound peak from a mass chromatogram acquired by an analyzer.

  A chromatograph mass spectrometer such as GC / MS or LC / MS can create a mass chromatogram based on the time course of the intensity signal of ions having a specific mass-to-charge ratio m / z. The waveform of this mass chromatogram shows the peaks derived from the ions of the target compound, the baseline of the mobile phase and solvent of the gas chromatograph and liquid chromatograph, that is, the baseline derived from the background ions, and other noises. It can be considered that three signals are superimposed.

  In order to identify or quantify the target compound based on these chromatograms, it is necessary to accurately separate the peaks, grasp the time position of the peak top, and obtain the peak height and area with high accuracy. is there. Conventionally, as a method for detecting a peak in a chromatogram, a change rate of signal intensity (amount of change in intensity per unit time) is generally obtained, and a peak start point or end point is determined based on the change rate. (For example, refer to Patent Documents 1 and 2).

  However, in general chromatograms, peak detection can be performed fairly accurately by the above method, but in mass chromatograms, peak detection using the above method is often difficult. That is, in the mass chromatogram, the waveform shape of the signal, particularly the base line shape, often varies greatly depending on the mass to charge ratio, and it is necessary to set an appropriate peak detection parameter for each mass to charge ratio. Therefore, when trying to perform peak detection processing on mass chromatograms for many mass-to-charge ratios, the operation is very complicated and troublesome. In addition, since there are many noises with waveform shapes similar to compound peaks, especially in mass chromatograms that contain a lot of noise, it is difficult to set appropriate peak detection parameters that separate these noises and peaks. It is.

Japanese Patent No. 3702931 JP-A-6-230001

  The present invention has been made to solve the above-mentioned problems, and the object of the present invention is to detect noise having a shape similar to a compound peak when there is a relatively large baseline fluctuation such as baseline drift. An object of the present invention is to provide a chromatogram data processing apparatus capable of detecting a peak with high accuracy without performing troublesome parameter setting even if included.

In order to solve the above-mentioned problems, the present invention provides a chromatogram obtained by a chromatograph including a separation unit for separating various components contained in a sample in a time direction and a detector for detecting the component-separated sample. A data processing device for detecting peaks from
a) baseline estimation means for obtaining a chromatogram baseline by performing a filtering process using a filter having a bandwidth and / or a cut-off frequency according to the degree of temporal variation of chromatogram data;
b) obtaining a threshold value considering the stochastic distribution of noise from the residual signal distribution of the chromatogram obtained by subtracting the baseline obtained by the baseline estimation means, and using the threshold value, Peak separation means;
It is characterized by having.

  The chromatogram data processing apparatus according to the present invention is particularly suitable for data processing of a mass chromatogram acquired by a chromatograph mass spectrometer whose detector is a mass spectrometer.

  For example, when processing mass chromatogram data, the baseline is basically composed of one or more low-frequency components in the mass chromatogram. When a large ion peak appears, the baseline of a mass chromatogram having a different mass-to-charge ratio may drop sharply. This is because the energy supplied for ionization in the ionization section of the mass spectrometer is almost constant, so that the amount of a certain substance rapidly increases in the ionization section and energy is consumed for ionization of the substance. Because of this, it is difficult to ionize other substances (including a solvent and a mobile phase) introduced into the ionization part almost simultaneously. Therefore, in order to cope with such a sudden drop in the baseline, the baseline estimation means uses a bandpass filter with a narrow bandwidth and / or a cut-off frequency in a portion where the degree of temporal fluctuation of the chromatogram data is steep. Apply a low-pass filter with lowering to find the baseline.

  Here, as a filter, a median filter for removing a peak so that the baseline is not pulled mainly due to the sudden drop as described above, and a smoothing filter that passes only the original low frequency component of the baseline. The low-pass filter can be used together. By such filtering processing, a baseline with drift can be estimated with high accuracy.

  Subtracting the baseline from the mass chromatogram leaves peaks and noise. Therefore, the noise / peak separation means predicts the distribution of the noise component from the distribution of the residual signal by a statistical method, and treats the peak that is not buried in the noise component. For example, when a mass spectrometer is used as a detector, the noise component is considered to follow a binomial distribution in which each molecule is ionized with a certain probability. Therefore, since the amplitude of noise is proportional to the square root of the baseline, signal normalization is performed using the proportional relationship. Then, in order to regard the normalized signal that is not buried in the noise component as a peak, the noise range is determined, and the noise and the peak are separated by a threshold value determined based on the range. Specifically, after normalizing the noise peak signal by dividing it by the square root of the baseline, in order to avoid the influence of the peak component, the standard deviation of the noise distribution is obtained using a signal near zero. Then, a value obtained by multiplying the standard deviation by a predetermined value that allows for a margin is defined as a threshold, and the peak and the noise are separated by this threshold.

  In addition, when it can be considered that there is no drift in the baseline, it is not necessary to perform the baseline estimation process as described above, and the processing time corresponding to that is wasted. Therefore, the chromatogram data processing apparatus according to the present invention further includes a drift determination unit that determines whether or not there is a baseline drift, and when it is determined that there is a baseline drift, the baseline estimation unit and the noise / peak separation are performed. It is preferable that peak detection using means is executed, and if it is determined that there is no baseline drift, an individual peak is determined from the aspect ratio of the peak on the chromatogram and peak detection is performed. By determining the presence or absence of baseline drift in this way and switching the peak detection method according to the result, peak detection processing accuracy is improved while shortening processing time by omitting unnecessary baseline estimation processing. be able to.

  According to the chromatogram data processing apparatus of the present invention, since the peak is detected by performing statistical processing on data having a baseline drift and containing a lot of noise, the waveform of the signal is simply obtained. The peak can be detected with higher accuracy than the conventional method of detecting the peak from the shape. In addition, it is not necessary to change the parameters to be set for peak detection for each mass-to-charge ratio, and mass chromatograms for all mass-to-charge ratios can be processed automatically. This eliminates the need for labor savings and improves throughput.

  The case where the chromatogram data processing apparatus according to the present invention is applied to a liquid chromatograph mass spectrometer (LC / MS) will be described as an example. FIG. 1 is an overall configuration diagram of the LC / MS.

  In the liquid chromatograph (LC) 1, the mobile phase stored in the mobile phase container 2 is sucked at a substantially constant flow rate by the liquid feed pump 3 and fed to the column 5. A sample to be analyzed is introduced into the mobile phase from the injector 4 at a predetermined timing, and is sent to the column 5 on the mobile phase. While passing through the column 5, various components contained in the sample are separated in the time direction and eluted in order from the column 5. The sample liquid containing the eluted sample components is introduced into the mass spectrometer 10 as a detector.

  In the mass spectrometer 10, the sample liquid is sprayed from an electrospray (ESI) nozzle 12 into the ionization chamber 11, which is a substantially atmospheric pressure atmosphere, whereby the component molecules in the sample liquid are ionized, and the generated ions are heated by a heating pipe. 13 is sent to the first intermediate vacuum chamber 14 which is a low vacuum atmosphere. In the ionization chamber 11, in addition to ESI, another atmospheric pressure ionization method such as atmospheric pressure chemical ionization (APCI) may be employed, or they may be used in combination. In any case, the ions are converged by the first ion lens 15 disposed in the first intermediate vacuum chamber 14, and sent to the second intermediate vacuum chamber 18, which is an intermediate vacuum atmosphere, through the orifice 17 at the top of the skimmer 16. While being converged by the second ion lens 19 disposed in the second intermediate vacuum chamber 18, it is fed into the analysis chamber 21 which is a high vacuum atmosphere.

  In the analysis chamber 21, only ions having a specific mass (strictly, mass to charge ratio m / z) pass through the space in the long axis direction of the quadrupole mass filter 22, and ions having other masses diverge midway. To do. The ions that have passed through the quadrupole mass filter 22 reach the ion detector 23, and the ion detector 23 outputs an ion intensity signal corresponding to the amount of ions that have reached. This output signal is digitized by the A / D converter 24 and input to the data processing unit 30. The data processing unit 30 is embodied by, for example, a personal computer in which dedicated control / processing software is installed, and includes a chromatogram creation unit 31, a peak detection processing unit 32, a qualitative / quantitative processing unit 33, and the like as functional blocks. .

  In general, the mass spectrometer 10 scans a predetermined mass range repeatedly and comprehensively detects ions in the mass range, while selecting only ions having a specific mass to charge ratio. By performing either or both of the selected ion monitoring measurements to be detected, chromatographic analysis is performed until a predetermined time elapses from the time when the sample is injected. Based on the data thus obtained, the chromatogram creation unit 31 creates a mass chromatogram and a total ion chromatogram, and the peak detection processing unit 32 performs peak detection based on the mass chromatogram data constituting the mass chromatogram. Execute. The qualitative / quantitative processing unit 33 identifies the compound based on the detected peak information and quantifies the identified compound.

  The LC / MS of the present embodiment is particularly characterized by the peak detection process executed by the peak detection processing unit 32. Next, this will be described in detail. FIG. 2 is a processing flowchart of the peak detection algorithm in this embodiment.

  First, the mass chromatogram data to be processed is acquired by, for example, reading from the storage unit (step S1). FIG. 4 is an example of a mass chromatogram. In general, a mass chromatogram has a waveform shape in which a low-frequency baseline and high-frequency noise are superimposed in addition to the peak derived from the target compound.

  A signal duration T having a signal intensity of 0 or more is calculated for the chromatogram data. If there are a plurality of signal durations, the maximum signal duration T is selected (step S2), and the maximum signal duration is selected. It is determined whether or not the width exceeds a predetermined value (step S3). For example, as shown in FIG. 5, when there is no baseline drift, the signal duration T is short, and when there is a baseline drift, the signal duration T is long. Therefore, the presence or absence of the baseline drift can be determined based on the magnitude of the signal duration width T. It should be noted that the predetermined value for determining the maximum signal duration time width may be changed by the user.

  If it is determined as Yes in step S3 and it is determined that there is a baseline drift, the following processes with baseline drift in steps S5 to S7 are executed (step S4). That is, first, the baseline is estimated by treating the baseline as noise data with a low frequency change (step S5). FIG. 3 shows the procedure of the baseline estimation process. In mass chromatogram data including noise, when many ions appear in other mass-to-charge ratios, the signal intensity may be drastically lowered due to the influence. This is because the total amount of energy supplied for ionization in the ionization chamber 11 is determined. Therefore, when a specific substance is introduced in a large amount (at a high concentration), a large amount of energy is required to ionize it. This is because the energy for ionization of other substances is relatively reduced and ionization efficiency is deteriorated. Therefore, in order to find a steep drop in the mass chromatogram data, a signal intensity change amount with the passage of time is calculated (step S51).

  FIG. 6 is a conceptual diagram for explaining a method of calculating the intensity change amount. As shown in the figure, a weighted average of an appropriate size is calculated for the changed portion of the signal in the first half and the second half, and the average values A and B are used, for example, (A−B) / (A + 1). The intensity change amount is obtained by the following formula. In order to perform weighting, a cosine window (Hanning window) centering on the boundary of the change portion of the signal can be used as a weighting window function. However, as long as the rate of change can be obtained from a certain signal width, another window function such as a Hamming window may be used, or a method other than the method using the window function may be used. Parameters such as the width of the window may be appropriately changed and set by the user.

  As described above, if the intensity change amount is obtained in each change portion of the signal on the chromatogram, the median (described later) is set so that the pass band becomes narrower as the intensity change amount increases, that is, the change becomes steeper. (Median value) The bandwidth of the filter is determined, and the cutoff frequency of the low-pass filter for smoothing is determined so that the lower the cutoff frequency is, the sharper the change is (step S52). Next, in order to prevent the baseline from being pulled by the peak component, a filtering process is performed on the chromatogram data using the median filter whose bandwidth is determined by the above process (step S53). Thereby, the peak is removed.

  Subsequently, a smoothing process is performed by the low-pass filter in which the cutoff frequency is determined as described above so as to pass the low frequency component of the baseline and cut off the higher frequency components (step S54). Here, a weighted average is used in a range of [−T / 2: T / 2] (T is the bandwidth obtained above) with 0.5 + 0.5 × cos (2πt / T) as a weight. . In addition to applying the low-pass filter using the cosine window as described above, another smoothing process capable of setting an appropriate pass band may be used. FIG. 7 is a diagram showing a result of performing baseline estimation by the above method. As shown in the figure, it can be seen that a baseline having a large drift can be extracted fairly accurately.

  Once the baseline reflecting the drift is obtained as described above, a noise threshold for separating the noise and the peak is calculated, and the peak is selected based on the noise threshold (step S6). . As a first step, only the noise and peak are extracted by subtracting the baseline estimated as described above from the mass chromatogram (noise peak value). FIG. 8A shows an example of such extracted signal waveforms, in which noise and compound-derived peaks are mixed.

  Next, the whole is normalized by dividing the amplitude of the noise peak value by the square root of the baseline. That is, when performing ionization with a mass spectrometer, each molecule is considered to follow a binomial distribution in which it is ionized with a certain probability, so that the noise amplitude is proportional to the square root of the baseline. Therefore, the noise peak value can be normalized as described above using this proportional relationship. FIG. 8B is an example in which the signal waveform of FIG. 8A is normalized, and it can be seen that the amplitude of noise is suppressed particularly in the latter half (C portion in FIG. 8B). By normalizing here, it becomes easy to obtain the probability distribution of the noise component.

  In such a normalized signal distribution, it is considered that a noise component occupies substantially within a range of, for example, ± 25% from the average value of the intensity. Therefore, as shown in FIG. 9 (b), assuming that the signal intensity is a normal distribution, the standard deviation within ± 25% (that is, within the 50% range) is multiplied by 3.65 to obtain the true value. A standard deviation σ is obtained. A value obtained by multiplying the true standard deviation σ by n is set as a threshold value, and a signal below this threshold value is regarded as noise, and a signal having a large value exceeding the threshold value is selected as a peak (see FIG. 8B). In this example, the noise range is ± 25% and the standard deviation magnification is 3.65. However, the present invention is not limited to this, and can be changed as long as a range that occupies most of the noise components can be set. is there. Further, if the noise distribution can be estimated in a state where the peaks are mixed, the noise peak can be separated using another method.

  Although the peak and the noise can be separated by the process of step S6, the peak range of each peak is determined based on the change in signal intensity before and after the peak thus obtained (step S7). Here, if the change in signal intensity is equal to or less than a predetermined minimum peak intensity, it may be ignored as a peak. Further, it is preferable to limit the peak range so that it does not extend to more than twice the time difference between the latest peak bottom and peak top, and basically use the peak bottom as the peak start point and end point.

  On the other hand, if it is determined No in step S3 and it is determined that there is no baseline drift, a process without baseline drift is executed (step S9). That is, a signal component having a signal intensity of 0 or more on the mass chromatogram is acquired as a compound peak (step S10). Here, when a signal intensity of 0 or more continues for a predetermined time or longer, it is regarded as not a single peak but a set of a plurality of peaks, and processing is performed so that the waveform is divided into a plurality of peaks, that is, each peak. FIG. 11 is an explanatory diagram of an example of the peak division method. Here, it is determined whether or not it is a peak division point by comparing the ratio t / h of the intensity difference h between the peak time width t and the peak top / peak bottom with a threshold value. When the ratio t / h is equal to or greater than the threshold value, it is regarded as a single peak without being divided. Thus, a peak range is obtained for each peak.

  As described above, a peak is detected regardless of the presence or absence of the baseline drift, but here, there may still be a mixture of noise determined as a peak due to erroneous detection. Therefore, noise removal by threshold processing is performed (step S8). For example, when a compound peak appears in another mass-to-charge ratio as described above, a sharp signal intensity change occurs due to the influence, and this appears as noise having a shape similar to the compound peak. Therefore, in order to exclude such false peaks, as shown in FIG. 10, the maximum value and the minimum value of the peak centered on the baseline within each peak range are obtained, and the ratio (Du / Dd) is calculated. It is then determined whether this is below a predetermined threshold. Here, the threshold can be set to 1, for example, but can also be determined as appropriate. If it is an original compound peak, it protrudes greatly above the baseline, so that a false peak due to the above cause can be eliminated. When the baseline is 0, since Dd = 0, there is no peak removed as noise.

  Further, as shown in FIG. 11, the case where the ratio between the peak height h and the time width t is smaller than a predetermined threshold value is regarded as noise and removed. Thus, when a compound peak is finally detected from the mass chromatogram, peak information such as the peak top position (time), peak top height, or peak area is obtained for each peak. -Delivered to the quantitative processing unit 33. The qualitative / quantitative processing unit 33 identifies a compound corresponding to this peak by, for example, checking the mass-to-charge ratio and peak information against a database.

  As described above, the LC / MS according to the present example detects a peak derived from a compound with high accuracy even when the baseline of a mass chromatogram has a drift or a lot of noise is included. Qualitative analysis and quantitative analysis can be performed.

  In the above description, the characteristic data processing of the present invention is performed on the mass chromatogram data. However, other chromatograms such as a total ion chromatogram created by a chromatograph mass spectrometer and a mass spectrometer are used. The present invention can also be applied to a chromatogram created by a gas chromatograph or a liquid chromatograph using a detector other than the above. However, for example, since the noise distribution in step S6 uses the statistical properties of the noise distribution accompanying ionization in the mass spectrometer, if the detector is different, the noise distribution corresponding to the detector is different. It is necessary to perform modification so as to determine the noise threshold using statistical properties.

  It should be noted that the above embodiment is an example of the present invention, and it is obvious that any modification, correction, or addition as appropriate within the spirit of the present invention is included in the scope of the claims of the present application.

The whole block diagram of an example of LC / MS to which the data processor concerning the present invention is applied. The processing flowchart of the peak detection algorithm in a present Example. The flowchart which shows the procedure of a baseline estimation process. The figure which shows an example of a mask tomogram. The figure for demonstrating how to obtain | require a signal duration. The conceptual diagram for demonstrating the calculation method of signal strength variation | change_quantity. The figure which shows a mass chromatogram signal and the baseline estimated from this. The figure which shows the noise peak waveform before and behind correction | amendment by normalization. The conceptual diagram for demonstrating the threshold value setting method by statistical noise distribution estimation. The conceptual diagram for demonstrating the technique of the false peak exclusion resulting from a sudden signal strength change. The conceptual diagram for demonstrating the technique of the peak division | segmentation and noise removal using a peak waveform shape.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Liquid chromatograph 2 ... Mobile phase container 3 ... Liquid feed pump 4 ... Injector 5 ... Column 10 ... Mass spectrometer 11 ... Ionization chamber 12 ... Electrospray nozzle 13 ... Heating pipe 14 ... 1st intermediate | middle vacuum chamber 15 ... 1st Ion lens 16 ... Skimmer 17 ... Orifice 18 ... Second intermediate vacuum chamber 19 ... Second ion lens 21 ... Analysis chamber 22 ... Quadrupole mass filter 23 ... Ion detector 24 ... A / D converter 30 ... Data processing unit 31 ... Chromatogram creation unit 32 ... Peak detection processing unit 33 ... Qualitative / quantitative processing unit

Claims (3)

A data processing device for detecting a peak from a chromatogram obtained by a chromatograph including a separation unit that separates various components contained in a sample in a time direction and a detector that detects a component-separated sample,
a) baseline estimation means for obtaining a chromatogram baseline by performing a filtering process using a filter having a bandwidth and / or a cut-off frequency according to the degree of temporal variation of chromatogram data;
b) obtaining a threshold value considering the stochastic distribution of noise from the residual signal distribution of the chromatogram obtained by subtracting the baseline obtained by the baseline estimation means, and using the threshold value, Peak separation means;
A chromatogram data processing apparatus comprising:   Drift detection means for determining the presence or absence of drift in the baseline of the chromatogram, and when it is determined that there is a baseline drift, peak detection using the baseline estimation means and noise peak separation means is performed; 2. The chromatogram data processing apparatus according to claim 1, wherein when it is determined that there is no baseline drift, peak detection is performed by determining individual peaks from the aspect ratio of the peaks on the chromatogram.   The chromatogram data processing apparatus according to claim 1 or 2, wherein the detector is a mass spectrometer, and the chromatogram is a mass chromatogram created by paying attention to a specific mass-to-charge ratio.