JP2006177980A - Chromatographic data processing device, chromatographic data processing method, and chromatographic analyzer - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a chromatographic device for readily performing curve fitting, using nonlinear least-squares method, even with respect to a chromatogram in which a plurality of overlapping peaks are present. <P>SOLUTION: In the chromatographic data processing device for processing chromatogram data, obtained by separating a sample to be measured by a column and detecting the separated sample, fitting processing is applied to each peak with respect to an arbitrary time region, in which a plurality of peaks are present in the chromatogram from a faster region or slower time region, and by subtracting processed peaks from the chromatogram in the arbitrary time region, the plurality of peaks of the chromatogram are decomposed. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a chromatographic technique such as a liquid chromatograph, and more particularly to a data processing method.

  In chromatography such as liquid chromatograph analyzers and gas chromatograph analyzers, a sample to be measured is separated for each component by passing through a column, and the separated components are separated by a chromatographic detector such as a photometer. The output value for each elapsed time is detected.

The signal output from the chromatograph detector is recorded as time series data having a time interval of about several hundred ms. This is a so-called chromatogram having the signal intensity on the vertical axis and the retention time on the horizontal axis. In general, the signal intensity is converted into a digital value I _j every predetermined time (time t _j ), and data processing is performed.

  FIG. 3 illustrates a chromatogram obtained by conducting biological fluid amino acid analysis.

As shown in FIG. 3, during the retention time of 23 to 34 (min), from Gly (glycine)
Eleven component peaks up to Tyr (tyrosine) are very densely present. In such a case, conventionally, area / quantitative calculation is performed using a vertical division method in which a perpendicular line is taken from a minimum point between peaks, so-called “valley”. However, this method produces an error of several tens of percent as the peak overlap is strong and produces inaccurate results. Therefore, in order to quantitatively analyze the chromatogram in such a state with higher accuracy, it is common to extend the analysis time and improve the resolution.

  Further, in order to perform quantitative calculation without extending the analysis time even when the overlap of peaks is strong, it has also been attempted to perform quantitative calculation in data processing using numerical analysis. For example, a method using a nonlinear least square method is used.

When the nonlinear least square method is used, at least three independent parameters (A: area, t _R : retention time, σ: standard deviation) are used as variables in order to fit a single component peak. Therefore, in order to fit the peaks of a plurality of components, it is necessary to obtain three parameters A _i , t _Ri and σ _i for the component (i), respectively.

As conventional examples using the non-linear least squares method, JP-A-6-324029 and JP-A-63-3
No. 151851 is published.

  In these examples, the overlapping peaks on the chromatogram are curve-fitted using a waveform function such as a Gaussian function or an EMG function (Exponentially Modified Gaussian) that can represent an asymmetric peak. It is disclosed. As shown in these examples, by using the nonlinear least square method, overlapping peaks can be decomposed into individual peak waveforms, and peak sizes such as peak areas corresponding to individual peak components can be obtained and quantified. Calculations can be made.

Japanese Patent Laid-Open No. 6-324029 JP-A-63-151851

  However, in the conventional example using the non-linear least square method, the decomposition of the peak waveform targets two or three overlapping peaks, and no further overlapping peaks are decomposed.

  Usually, when performing peak waveform decomposition by nonlinear least squares curve fitting, the more the number of peak components is 3, 4, 5,..., The more difficult the fitting process converges or the peak This is because a phenomenon occurs in which the decomposition is not performed correctly (the error increases).

  For example, in the case of the chromatogram of FIG. 3, if 11 components from Gly to Tyr are to be fitted at a time, 33 parameters of 11 × 3 must be determined at a time. This is a very difficult calculation process for the non-linear least-squares method, and various ideas are required to solve it.

  Therefore, when there are a large number of overlapping peaks on the chromatograph, if the curve fitting is performed using the nonlinear least square method, the measurer will be able to converge for a few peaks one by one. Window) must be specified, which is very time consuming and the area is determined by the hand of the person, so there is a problem in the reliability of the calculation results.

  An object of the present invention is to provide a chromatographic apparatus capable of easily performing curve fitting using a nonlinear least square method even for a chromatogram having a plurality of overlapping peaks.

  In order to achieve the above object, the present invention is characterized in that a sample to be measured is separated by a column, and a chromatogram data processing apparatus for performing data processing of a chromatogram obtained by detecting the separated sample, For an arbitrary time domain in which multiple peaks exist, a fitting process is performed for each peak from the earlier of the time domain or the later of the time domain, and the peak subjected to the process is included in the arbitrary time domain. By subtracting from the chromatogram, the peaks of the chromatogram are resolved.

  According to the present invention, a peak can be easily decomposed only by defining some setting conditions for a chromatogram having a plurality of overlapping peaks, particularly, an overlapping peak of three or more components. Thereby, the precision of quantitative analysis and qualitative analysis can be improved. It also improves the ability to determine the baseline. Furthermore, the labor of the measurer for data processing can be greatly reduced.

  In addition, the subject, an effect | action, and effect of this invention are demonstrated in detail in the column of the form for implementing invention mentioned later.

  Examples of the present invention will be described below.

  FIG. 2 is a schematic diagram of a liquid chromatograph apparatus to which the present invention is applied. First, the eluent 1 is sent to the column 3 by the liquid feed pump 2 in accordance with an instruction from the control unit 5. A sample supply unit 8 is provided between the liquid feed pump 2 and the column 3, and the sample is supplied to the eluent according to an instruction from the control unit 5 from the sampler 7 provided with the sample. The sample is separated by the column 3 and detected by a detector 4 such as a UV detector. The chromatogram which is the detected data is sent to the control unit 5 for data analysis, and the result is displayed on the display 6 or printed by the printer 9.

  Next, data processing in the control unit 95 when the chromatogram of FIG. 3 is obtained as a result of detection will be described.

  In the present embodiment, chromatogram data processing is mainly performed according to the following procedure.

Step 1: Specifying the time interval for fitting Step 2: Selecting the weighting pattern Step 3: Selecting the fitting direction Step 4: Clicking the fitting execution button Step 5: Displaying the result display Figure 1 shows the data processing of the above chromatogram The detailed process is shown. In this embodiment, an example in which fitting processing is performed for each two components will be described.

  First, a chromatogram to be fitted is selected from the obtained detection results, and the process is started (100). Here, it is assumed that the variable i = 1.

  Next, a holding time interval (time window) to be subjected to the fitting process is set (101). Here, in the chromatogram of FIG. 4, a peak group of 11 components from Gly to Tyr is set as a target.

  As shown in FIG. 4, the time window is set by dragging the time axis of the chromatogram displayed on the display 96 with a cursor. Alternatively, the peaks Gly and Tyr of the displayed chromatogram can be picked to select the fitting start peak and end peak. In addition, it is possible to directly input the start and end times.

  Next, a waveform function used for fitting calculation is set. For example, “Gaussian” or “EMG” is selected as the waveform function. This selection is performed by selecting a waveform function to be used on the display 96 from the dialog box shown in FIG.

  In this embodiment, two components are fitted, but in order to more accurately determine the front peak in the fitting direction, that is, the waveform of the first peak, the influence of the signal after the waveform of the second peak is reduced. There is a need to. Therefore, next, the weighting function as shown in FIG. 6A is set (103).

The weighting function shows the weighting for each of the two peaks. FIG.
In the example of (a), the weight w = 1 from the peak start time 21 to the peak end (valley) time 23 of the first peak, and the peak end (valley) time 25 from the peak start (valley) time 23 of the second peak. Up to here is an example in which a linear gradient is added to the weight w and dropped from 1 to 0. A variety of weighting functions can be used, and examples are given below.

FIG. 6 (b): Curved gradient. FIG. 6C: S-shaped curve gradient. FIG.
(D): A linear gradient from the first peak start point to the second peak end point. FIG. 6 (e): the second peak end point multiplied by 0.5. FIG. 6F: a linear gradient from the second peak maximum point to the second peak end point. FIG. 6 (g): Emphasis on the left half of the first peak, and a linear gradient from the first peak maximum point to the second peak end point. The weight is multiplied by 0.5 at the first peak end point (second peak start point). FIG. 6 (h): A weight of 0.5 is applied flatly from the first peak end point (second peak start point) to the second peak end point. FIG. 6 (i): equivalent when weighting is not performed.

  Specifically, the dialog box shown in FIG. 7 is displayed on the display 96 and set by selecting a weighting pattern. Here, by selecting an arbitrary setting and pressing an “option” button, the above-described various pattern function graphs shown in FIG. 6 can be set. The various patterns in FIG. 6 are stored in advance in a storage device such as a hard disk in the control unit 5, and can be easily read out from the storage device and used for processing by being specified in the dialog box in FIG. .

  Next, the direction in which the fitting calculation is performed is determined (104).

  That is, in the example of FIG. 3, it is determined whether the process is performed from the front (Gly side) or the rear (Tyr side). Specifically, the setting is performed in the dialog box of FIG.

  The following describes an example in which processing is performed from the Gly side.

  This completes the setting of the conditions necessary for the fitting process.

  Therefore, when the “OK” button is selected in the dialog box of FIG. 7, a dialog box as shown in FIG. 8 is displayed, and the calculation process of fitting is started by pressing the “execute” button.

  When the calculation process is started, first, the number of peaks in the set time window is detected and set to Imax (106).

  Specifically, the inflection point within the set time window is detected, the maximum point and the minimum point are obtained, the maximum point is the peak apex, and the minimum point is the peak end point and start point. Define the interval. The peak valleys (minimum points) obtained at this time are stored in association with the retention times at which the points exist.

  Next, the difference between Imax and i is obtained (107). If the difference is 3 or more components, the process proceeds to step 107.

  When it is determined in step 107 that the difference is 3 or more, the width of the weighting function set in step 103 is adjusted in accordance with the peak valley detected in step 105 (108).

  FIG. 9 shows an image in which an area for the first fitting process is set. The width of the weighting function is adjusted in accordance with the width of the fitting area as shown in FIG.

  Next, a fitting process is performed on the two front components in the fitting direction (109).

  Below, the specific process of fitting is demonstrated.

  As described above, Gaussian or EMG is used for the waveform function f (t) for fitting. First, Formula 1 shows an example of using Gaussian.

Here, at + b represents a baseline. In this embodiment, since the fitting process is performed for each of two components, Equation 1 is obtained. However, if the ability of the nonlinear least square method can be ensured, it can be increased to three components. Of course, expansion to EMG is also possible. Formula 2 shows an example of replacing one g _i (t) of Formula 1 with EMG.

In the least square method, the fitting parameters A _i , t _Ri , σ _i , and τ _i are determined so as to minimize S ₁ of the following Expression 3 or S ₂ of Expression 4. Here, I _j is the signal intensity of the measured chromatogram, j is a subscript representing the time t _j , and N is the number of data points in the time interval. Any of S ₁ and S ₂ may be used.

The first component waveform is determined (110) when the smaller t _R in f (t) of Equation 1 obtained here (that is, the earlier holding time) g _i (t) corresponds to the first peak Gly (110).

Next, the first component waveform g ₁ (t) is subtracted from the original measurement chromatogram, and a chromatogram obtained by cutting out Gly is created (111). FIG. 10 shows an image of a chromatogram obtained by cutting out Gly. Note that it is preferable not to subtract the baseline at + b because it does not affect the peak waveform.

  Next, the variable i is incremented and the process returns to Step 107.

  Thereafter, fitting is performed on Ala and Cit by the same procedure as that for fitting the first peak, and the second component waveform corresponding to the second peak Ala is determined.

  After that, until the two components Leu (leucine) and Tyr at the rear end are left, Ala, Cit... And subtraction are sequentially performed, and when the remaining two components are obtained (i = 2), The process proceeds to step 112, in which two-component peak fitting is performed without weighting, and the final two-component peak waveform is determined (114). Then, the peak waveform of the two components is subtracted from the original chromatogram (115), and the process is terminated.

  Through the above series of processing, all peak waveforms in the time window are decomposed. Further, according to the above processing, the remaining chromatogram from which all the peak waveforms have been extracted becomes the baseline, and the baseline is also obtained at the same time.

  When the above processing is completed, quantitative calculation (for example, concentration) is performed using the parameters for each peak obtained by the above processing. Then, as a calculation result, the following display as shown in FIG. 11 is made on the display 96 for each peak (each component).

  Here, since the uncertainty (variation) is obtained for the parameters for each component by the above processing, the uncertainty can be calculated for the quantitative value based on this uncertainty.

  A feature of this embodiment is that the uncertainty of the quantitative value can be calculated. The calculation method for determining the uncertainty of the quantitative value follows the error propagation equation.

  In addition, the superiority or inferiority of the fitting can be determined by the magnitude of the uncertainty of the quantitative value (it can be said that the smaller the uncertainty, the better). For example, this uncertainty can be used to determine the superiority or inferiority of the calculation result of overlapping portions when fitting is performed from both front and rear ends, which will be described later.

In the present invention, the above-described processing for performing peak decomposition while performing peak extraction is called “sequencing”, and a chromatographic data processing apparatus that performs this sequencing is called “chromatographic peak sequencer”.

Although the above description is the fitting calculation from the front end, it can also be executed from the rear end by setting the fitting direction in the dialog box of FIG. In this case, g ₂ (t) indicating the second peak is subtracted from the chromatogram, and the shape of the weighting function is reversed left and right. In this order, in the case of FIG. 4, it can be cut out in the order of Tyr, Leu, Ile (isoleucine) from the one having the longest retention time.

  Ideally, the fitting may be performed from either the front or rear, but in practice, it is desirable to perform the fitting from both ends to the strongest overlapping peak. When performing such fitting, “automatic determination” is selected in the dialog box of FIG.

  For example, in the chromatogram of FIG. 3, since Lue and Ile have strong overlap, forward sequencing is performed from Gly to Cysthi (cystathionine) or Ile, and backward sequencing is performed to Lue or Ile. In this way, with regard to Lue and Ile, which have a strong overlap, it is possible to determine the suitability of fitting from both the front and rear, and to adopt and determine either fitting result.

Here, the uncertainties (variations) of the fitting parameters A _i , t _Ri , and σ _i themselves are used as indices for determining suitability, or the statistical indices χ ² and the convergence rate of χ ² are used. Judgment can be made by using the number of calculation iterations watching If a combination of these parameters is used for determination, a more comprehensive determination can be made. In any case, for the purpose of quantitative calculation, the peak area value A _i needs to be obtained accurately, and this uncertainty should be emphasized.
(Another example of weighting function setting)
The weighting function set in step 102 is set on the dialog box. In addition to this, the table format program setting shown in Table 1 or the weighting function is overlaid on the chromatogram shown in FIG. It can also be set by graphical settings that display the settings together.

  The setting program in Table 1 shows an example in which FIG. 6G is set as an example. When the measurer inputs “weight value” and “shape”, the weighting function is defined. This Table 1 is displayed on the display.

FIG. 12 shows a method of setting while displaying the weighting function superimposed on the gamma tomogram on the display. For example, the cursor is moved around the nod (node) 31 by a pointing device and clicked to display the first peak start point of the feature point. Now, the noud 31 is confirmed. If it is desired to move the nod 31, so-called dragging, which is released after picking, is also possible. Next, click around the nod 32 and drag to a point that is perpendicular to the nod 31 and has a weight of 1 to confirm the nod 32. Here, a vertical line from the nod 31 to the nod 32 is automatically drawn and determined. Similarly, the nodes 33 and 34 can be determined and a weighting function can be set. If you want to change the weighting function, click on any node, highlight it, and change it. You can also delete the noise. You can also add a new node by clicking anywhere. If you want to make a line between the nodules curved, click the target line, highlight it (change the display color), and right-click to change the attribute. Here, you can specify concave, convex, etc.
(Another embodiment of sequencing)
In the above-described sequencing example, the measurement is performed once and a chromatogram is obtained and data processing is performed for a while. However, the sequencing process can be performed simultaneously with the measurement of the sample. The embodiment in that case will be described below.

  The sequencing process at the same time as the measurement proceeds as follows.

Step 1: Setting the time interval (time window) for fitting Step 2: Setting the weighting function Step 3: Setting the fitting direction Step 4: Measuring the sample Step 5: Fitting process Step 6: Displaying the results Steps 1-3 Then, as shown in Table 2, the time program is set in advance before measurement. The time window is set by setting the time from the start of fitting to the end of fitting. In addition, the time window can be set by selecting a fitting start peak and an end peak such as “Gly” to “Tyr” as peak names.

  As described above, according to the chromatographic peak sequencer of the present invention, the fitting of a chromatogram in which a plurality of peaks overlap can be automatically executed only by setting some conditions.

  At the same time, a highly accurate baseline can be calculated. Also from the aspect of having a highly accurate baseline determination function, the chromatographic peak sequencer of the present invention is considered to be a good data processor. However, since fitting is performed assuming that the peak waveform is Gaussian or EMG, care must be taken when the actual peak waveform deviates significantly from these analytical function forms. In such a case, it is necessary to obtain an isolated peak waveform from a standard sample for each component and register a specific waveform function in advance.

It is a flowchart figure which shows the fitting process of this invention. It is a schematic block diagram of a chromatograph apparatus. It is a 70 minute chromatogram of the biological fluid amino acid analysis method. It is the example of a display of the chromatogram which set the time window. It is a display example of a dialog box for setting a waveform function. It is an example of a weighting function. It is a display example of a dialog box for selecting a weighting function and a fitting direction. It is a display example of the dialog box which designates execution of a fitting process. It is an image figure in which the first fitting area was set. It is an image figure of the chromatogram from which 1 component was cut out. It is a display example of a fitting result. It is a figure which shows the example of graphical setting of a weighting function.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Eluent, 2 ... Liquid feed pump, 3 ... Column, 4 ... Detector, 5,95 ... Control part, 6,96 ... Display, 7 ... Sampler, 8 ... Sample supply part, 9 ... Printer, 21 ... Peak Start time, 23, 25 ... Peak end (valley) time, 31, 32, 33, 34 ... Noud.

Claims (12)

In a chromatographic data processing apparatus for performing data processing of a chromatogram obtained by detecting a sample to be measured by a column and detecting the separated sample,
For any time region where a plurality of peaks of the chromatogram are present, a fitting process is performed for each peak from the earlier of the time region or the later of the time region, and the peak subjected to the processing is subjected to the arbitrary time A chromatographic data processing apparatus, wherein a plurality of peaks of a chromatogram are decomposed by subtracting from a chromatogram in a region. In claim 1,
The chromatographic data processing apparatus characterized in that the fitting process calculates at least an area, a holding time, and a standard deviation. In claim 1,
2. The chromatographic data processing apparatus according to claim 1, wherein the fitting process starts from both ends of an earlier time and a later time in an arbitrary time domain. In claim 1,
A chromatographic data processing apparatus characterized in that the remaining data obtained by subtracting all the peaks in the chromatogram is used as a baseline. In a chromatogram data processing method obtained by detecting a sample to be measured by a column and detecting the separated sample,
Designating any region where there are multiple peaks of the chromatogram;
Performing a fitting process for each peak from the front end or the rear end on the chromatogram of the designated region;
Subtracting the peak subjected to the fitting process from the chromatogram in the arbitrary region, and a chromatographic data processing method. In claim 5,
The chromatographic data processing method, wherein the fitting process calculates at least an area, a holding time, and a standard deviation. In claim 5,
A chromatographic data processing method comprising the steps of: setting a weighting function used for fitting processing; and setting the direction of the fitting processing. Sample separation means for separating the sample to be measured for each component, detection means for detecting the sample separated by the sample separation means, data processing means for obtaining a detection result of the detection means and performing data processing, and data processing In a chromatograph analyzer comprising display means for displaying results,
The data processing means includes means for designating an arbitrary holding time region in which at least three component peaks of a chromatogram obtained as a detection result of the detecting means exist, and a front end or a rear of the designated holding time region. A means for designating from which end the fitting process is performed and a means for designating a weighting function and a waveform function used in the fitting process, and displaying the fitting process result for each peak on the display means A chromatographic analyzer characterized by the above. In claim 8,
The chromatographic analyzer characterized in that the fitting process calculates at least an area, a holding time, and a standard deviation. In claim 8,
The chromatographic analyzer characterized in that the data processing means subtracts a peak for which the fitting process has been completed from a chromatogram in the arbitrary holding time region. In claim 8,
The chromatograph analyzer is characterized in that designation of the arbitrary holding time region, designation of the direction of the fitting process, and designation of the weighting function are performed by a dialog box displayed on the display means. In claim 11,
The weighting function is superimposed and displayed together with the chromatogram on the display means, and can be arbitrarily changed.

Images