WO2005121776A1

WO2005121776A1 - Method and device for the automatic development of models of chromatographic, electrochomatographic and electrophoretic methods

Info

Publication number: WO2005121776A1
Application number: PCT/DE2005/001046
Authority: WO
Inventors: Imre Molnar; H. J. Rieger; Dimas Fernandez De Souza; Robert Albrecht; Thomas Jupille
Original assignee: Imre Molnar; Rieger H J; Dimas Fernandez De Souza; Robert Albrecht; Thomas Jupille
Priority date: 2004-06-11
Filing date: 2005-06-13
Publication date: 2005-12-22
Also published as: DE112005001959A5; EP1754055A1

Abstract

The invention relates to a method for the automatic development of models of chromatographic, electrochomatographic and electrophoretic measurements, especially in high pressure liquid chromatography (HPLC), capillary electrophoresis (CE) and capillary electrochromatography (CEC). The inventive method is characterized by a) separation of at least one substance mixture of at least one sample on at least one column or capillary into individual components and detection of the separated components after a component-specific retention time tR on the column or capillary using a component-specific signal intensity S in the form of a detector signal, whereby at least one parameter (x, ..), preferably two parameters (x, y, ) of the measurement conditions are automatically varied; b) automatic generation of a first data set for determining the functional correlation of the signal intensity S of the separated components and the retention time tR in the form of component-specific peaks, whereby the at least first data set is associated with the at least one parameter (x, ..), preferably two parameters (x, y, ) of the measurement conditions; c) automatic evaluation of the functional correlation of the distance of the component-specific peaks in the form of the critical dissolution and the at least one parameter (x, ..), preferably two parameters (x, y, ) of the measurement conditions, and generation of at least one second data set for establishing a virtual separation model based on the functional correlation so determined; and d) generation of at least one third data set for determining a parameter of the method robustness Rb. The invention also relates to a device for carrying out said method.

Description

Method and device for automatic model development of chromatographic, electrochromatographic and electrophoretic methods

The invention relates to a method according to the preamble of claim 1 and a device according to the preamble of claim 23.

Chromatographic methods such as gas chromatography or liquid chromatography are proven methods for the qualitative and quantitative analysis of complex substance mixtures.

L O In gas chromatography, the chromatographic selectivity can practically not be influenced by the mobile phase. Separation optimization is therefore carried out using different separation columns. This process is discontinuous, i.e. no continuous changes in selectivity can be achieved with the separation columns.

L 5 However, the forces between eluent molecules play a major role in all liquid chromatography processes. The influence of the properties of water on the continuous change in the chromatographic selectivity is particularly pronounced in aqueous solutions. On this possibility of continuously changing the

20 selectivity, the present method is turned off.

Chromatographic, electrochromatographic and electrophoretic methods are tried and tested methods in aqueous or aqueous-organic eluents to carry out qualitative and quantitative analyzes of complex substance mixtures.

25 High-pressure liquid chromatography (HPLC) with so-called “reverse phases”, in English “reversed phase chromatography (RPC)”, has proven to be robust and reliable for the quantitative analysis of substance mixtures in chemical, pharmaceutical and life sciences. The RPC is based on the “Solvophobe

30 theory "by Cs.Horvέth, W.Melander and I.Molnar, (J.Chromatogr., 1976, 125, 129), on an interaction of the components or substances to be analyzed with both a stationary phase and a mobile phase. Usually the mixture of substances to be analyzed is applied to a column with a specific column material and eluted with an eluent, whereby the surface of the column material can be hydrophobic, for example

35 or ionic character, whereby specific hydrophobic, polar or ionic interactions of the individual components of the mixture of substances with the column material are brought about. The interactions of the individual components naturally vary, so that they are eluted at different times with a component-specific retention time, as a result of which the mixture of substances is separated into its Individual components can be made. In capillary electrochromatography (CEC) or in capillary electrophoresis (CE), an electrical voltage is additionally applied to the separation columns to increase the selectivity. The individual components are detected by one of their specific physical properties, for example their absorption at a specific wavelength, or their molar mass, and represented as individual peaks in a diagram or chromatogram.

Reliable and reproducible quantitative analyzes of substances and mixtures of substances using HPLC are extremely important in quality control, especially of pharmaceutical, cosmetic and food industry products. Toxic substances must be discovered and eliminated in good time (see Contergan), other, less dangerous ones must not exceed a limit in their concentration. In practice, reliable reproducibility of the analytical results at any time and anywhere is crucial. The quality of the products is closely related to the quality of the analytical processes. It is therefore important to automatically develop the chromatographic quality of the HPLC, CEC and CE methods and to define them using suitable processes.

However, reliable reproducibility is often not guaranteed with the methods previously used in practice, since these often yield different and sometimes even surprising results. Different retention times and values of the critical resolution are generally observed, the term critical resolution being understood to mean the distance between the two narrowest peaks, each defining a single component. The critical resolution represents an important criterion in the quality control of pharmaceutical products, and may be used within the scope of

Routine measurements should not fall below a previously defined limit.

It is common practice to determine suitable optimal analysis conditions to carry out a large number of experiments using the "trial & error" method, in which the composition of the stationary phase (column material) and the mobile phase (eluent) are varied. A strong influence on the separation quality - and thus on the product quality - also exert the external conditions such as temperature or runtime. During the application period of an analytical method, adjustments must also be made in order to comply with the so-called "system suitability test" (SST), which measures the quality of the methods to be developed.

However, this approach of this non-systematic method development, which has been common until now, means a high material consumption of solvents, separation columns, samples and a high time requirement. Furthermore leads to non-systematic method development Methods without documented quality. Transferring a measurement method to other systems and conditions is easier if the method is reliable and understandable.

The invention is therefore based on the problem of providing a method which systematically carries out systematic method development for chromatographic, electrochromatographic and electrophoretic measurements in aqueous-organic eluents, facilitates the assignment of the components, and the transfer of the data into a simulation software for HPLC, CEC or CE takes place and thus allows a reliable reproduction of the developed methods regardless of location.

This object is achieved according to the invention by a method with the features of claim 1.

The method according to the invention for automatic model development of chromatographic, electrochromatographic or electrophoretic measurements, in particular in high-pressure liquid chromatography (HPLC), capillary electophoresis (CE) or capillary electrochromatography (CEC), is characterized in that in a first step a) at least one separation of at least one substance mixture a sample is carried out on at least one column or capillary in the respective individual components. The separated components are detected at the column or capillary outlet after a component-specific retention time t _R by means of a component-specific signal intensity S in the form of a detector signal. The separation takes place under various external conditions influencing the separation quality, at least one parameter (x, .-.) _> Preferably two parameters (x, y, ...) of these measurement conditions being changed automatically.

In a second step b), at least one first data set is automatically generated to determine the functional relationship between signal intensity S of the separated components and the retention time t _R in the form of component-specific peaks, the at least first data set corresponding to the at least one parameter (x, ... ), preferably two parameters (x, y, ...) is assigned. A data record can include data from at least one or more separation runs.

In a third step c), the functional relationship of the distance between the component-specific peaks in the form of the critical resolution and the at least one parameter (x, ...), preferably two parameters (x, y, ...), of the measurement conditions is automatically evaluated , and generates at least one second data set with which a virtual separation model based on the determined functional relationship.

In a fourth, final step, a third data set for determining a parameter of the method robustness R _{b is} generated.

Robustness R _b is understood here to mean that the result of the calculation of the critical resolution changes only slightly or not at all in the event of changes in the working conditions within parameter-dependent limits which are determined from the maps of the critical resolution. A simple connection that suggests high robustness would be, for example, a numerical evaluation of the first derivatives at the individual data points. If the first derivatives remain small, there are no major jumps in the original data if you analyze the neighboring points from one point. In principle, however, other robustness measures are also suitable.

The method according to the invention is thus based on a few experimentally measured parameters of the signal intensities S and retention times t _R , during the measurement of which a certain number of parameters (x, y, ...) influencing the measurement conditions are changed automatically, while other parameters which the result

0 influence can be kept constant. This allows the determination of the running behavior or retention behavior of each peak in at least one sample depending on the at least one automatically changeable parameter (x, ...) based on the experimental measurement data.

: 5 The method according to the invention thus enables the determination of virtual separating runs for each value of an experimentally adjustable parameter. The conditions of the separation runs can be virtually optimized with a view to maximizing the critical resolution while minimizing the time duration of an analysis process. Extensive, time- and material-intensive experiments are no longer necessary. The separation of the substance mixture is preferably carried out at at least two different parameter points of the at least one parameter (x, ...), preferably at least two parameters (x, y, ...) , so that at least (2x, ...), preferably at least [2 (x «y), ....], different separation runs with different data records are generated. 5 The at least one parameter (x, ...), preferably two parameters (x, y, ...) of the external measurement conditions is advantageously selected from the group of the gradient time tG, temperature T, pH of the eluent, composition of the eluent aqueous and organic phase, composition of the organic phase, ionic strength of the aqueous phase, concentration of additives, for example SDS.

Columns for reversed-phase chromatography, such as RP-C8 or RP-C18 columns, also with chemically bound ligands that contain polar functional groups, or CE or CEC capillaries with any filling material are preferably used as chromatographic columns.

It is also advantageous to determine the component-specific signal intensity S by measuring the light attenuation at a specific wavelength in a wavelength range from 220 to 800 nm of a detector, by measuring the fluorescence or the molecular weight and / or by another type of detection.

The value of the specific signal intensity S is determined by the peak area A as the product of the elution volume [mL] and the concentration [μg / mL] of the substance, the peak area A being directly proportional to the quantitative mass [μg] of the individual component. The prerequisite for this is the application of the same sample in injection volumes of the same size.

The assignment of the at least first data record to the at least one parameter (x, ...) is illustrated by means of at least (x, ...) diagrams and (x, ...) tables. When using at least two parameters (x, y, ...), the assignment is advantageously carried out using at least [(x ^* y), ...] diagrams and [(x ^* y), ...] tables.

The diagrams created are advantageously chromatograms or electropherograms. The values of the peak areas A displayed in the diagrams, preferably at least two chromatograms, are advantageously plotted via the peaks. For better clarity, the displayed values are advantageously simultaneously reduced by a factor calculated by the software, preferably by a factor of 100,000. Furthermore, small peaks and / or peaks from early eluting

Components are eliminated. The displayed values of the peak areas are also reduced in the peak area table, preferably at least two peak area tables, by a factor calculated by the software.

After creating the diagrams, each peak defining an individual component is localized in the at least one (x, ...) diagram, preferably [(x * y), ...] diagrams. The functional relationship between retention time t _R and peak area A results in a for each peak

Data record generated and in at least one (x, ...) peak table, preferably [(x-y)]

Peak tables, summarized. The localization of a peak defining a single component in the at least (x, ...), preferably [(x ^* y), ...], diagrams is advantageously carried out by considering the standard deviation of the corresponding peak areas A in the at least (x,. ..), 5 preferred [(x ^* y), ..] peak tables.

Standard deviation here means the deviation of the peaks or peak areas from one another in the experimentally determined diagrams. In the diagrams, these pecas each define the same individual component, but may have different elution behavior due to the different measurement conditions.

In this context, a standard deviation of less than or equal to 10% means a correct assignment of the peaks defining an individual component in all diagrams. L5 If the standard deviation has a value of 10 to 60%, a careful examination of the peaks is necessary. A standard deviation of greater than or equal to 60% indicates an incorrect assignment of the peaks.

The values with a standard deviation of less than or equal to 10% are highlighted with a green color in 20 of the peak table and the values with a standard deviation of more than 60% are highlighted with a red color. For the assignment of the peaks by means of the standard deviation, the prerequisite for using the same sample mixture and for giving up the same injection volumes in the individual experiments must be fulfilled. 5 Correct allocation of peaks with a standard deviation of over 10% must therefore be done manually in the peak table.

The virtual separation model advantageously includes a data record with a clear data assignment of the critical resolution to the corresponding parameters of the 0 signal intensities S and retention times t _{R of} a virtual separation run. The critical resolution R _s is determined by at least one parameter (x, ...), preferably at least two parameters (x, y, ...) of the measurement conditions.

The critical resolution R _s is determined by the distance between the peaks that define a single component between the two narrowest pecas, 5. The critical resolution is calculated mathematically using the equation _s = 2 (t _R2 - t _Rl ) / w ₁ + w ₂ (1) where t _R1 and t _{R2 are} the retention times of two different individual components, the peaks of these individual components being adjacent in the chromatogram or diagram, and w, and w _{2 being} the peak width at the base of the respective peaks.

It is also advantageous if, for any value of the at least one parameter (x, ...), preferably at least two parameters (x, y, ...), at least one virtual separation run from the data record of the critical resolution R _s can be determined.

The method also enables a comparison of the experimental and virtual separation runs to check the agreement or model validation. An optimal separation run can thus be modeled for any value of the parameters x and y chosen as desired.

Furthermore, information about column dimensions such as length and inner diameter, column material, grain diameter and flow rate as well as the gradient of the eluent can be changed. The gradient of the eluent can be changed automatically with at least 3 or more segments.

An automatic three-stage gradient can be calculated or modeled for a separation run. An optimal linear gradient is preferably calculated by inserting at least two additional gradient points, specifying the maximum runtime (max time), specifying the step size with which the two different gradient points are changed along the time axis (time increment), specifying the step size, with which changes the two different gradient points along the axis with the percentage of component B of the eluent (concentration increment) and the flow rate of a run (flow).

Thereby a) the critical resolution is maximized, b) the shortest analysis time is sought or determined, and c) after the flow rate (Flow) has been specified, the shortening of the analysis time under

Calculation of the changed selectivity, i.e. changed critical resolution of a run, modeled.

The method robustness is determined from the critical resolution map, which provides specific information about the values within which the experimental parameters may fluctuate in order to maintain a predetermined critical resolution. At least three resolution values are read from the map of the critical resolution. The first value is the experimentally determined parameter (nominal value). The second and third values determine the upper and lower limits of the specified fluctuation ranges. A percentage value is calculated from the first and the smaller of the second and third values, which indicates the maximum deviation of the critical resolution from the critical resolution at the nominal value. Furthermore, a minimum resolution is specified from which the permitted fluctuation ranges are determined. 5 Ultimately, the absolute value of the specified resolution is decisive for the applicability of a measurement method. The absolute value of the critical resolution can be taken into account very stringently by marking a method that falls below this value within the specified fluctuation ranges as "not robust". 0 Alternatively, the percentage value can be weighted. The weighting factor is based on whether and how In contrast to this, the term robustness essentially means how much the resolution value changes per unit of an experimental parameter 5 The method robustness R is advantageously used for at least 1 to 8 parameters

R _b = 100 (b / a) (Rswp / Rs _R ef) (2)

: o determined, where a and b is the level of the critical resolution, preferably b is less than a, R _{Sw is} the critical resolution at the potential working point and R _{S ef is} the reference value of the critical resolution, the permitted fluctuation ranges of the parameters using suitable algorithms per Parameters can be determined individually. R _sRe f is preferably _1.5 . 5 The percentage values of the individual robustness values can be mathematically linked and expressed in a special combination and can be displayed in a suitable ratio to a specified critical resolution value. The method according to the invention offers various advantages in the automatic model development of analytical measurement methods, in particular of chromatographic, electrochromatographic or electrophoretic measurements. 5 On the one hand, it is now possible to determine the effects of changing separation conditions on an existing method without having to experiment experimentally in addition to the systematic basic runs. The change in resolution for individual, several or all peaks under changed separation conditions can be followed. The effects, for example, of changed column dimensions, pH values, Eluent composition, temperatures or flow rates on a separation run can be determined and the run times for a higher sample throughput may be shortened. In particular, those experiments that are not optimal can be assessed and eliminated for their quality before they are carried out. This saves expensive resources and saves the cost of experiments that are not useful, the proportion of which is often over 90%.

The method according to the invention provides information about the robustness and reproducibility of a developed method. In addition, the method offers the possibility of transferring methods from one device to another, even across manufacturers, and thus standardizing the method for a large number of chromatographic or electrophoretic measurements with high precision between simulation and experiment, regardless of location.

A further advantage of the method according to the invention is the possibility that after a reference run according to the image section created there, the time axis sections are calculated and displayed graphically at the same time with all other chromatograms.

The following working methods are advantageously set up: -% B with 2-16 runs - tG with 2-16 runs

- pH, concentration of additives, with 3-16 runs each

- tG vs temperature (with 4-16 runs) - tG vs pH (with 6-16 runs)

- tG vs ternary eluent composition (with 6-16 runs)

- tG vs concentration of additives of the eluent (buffer, ion pairing agent, etc.) (with 6-16 runs)

It is also advantageous that saving the DryLab models as * .inp files is no longer necessary, since the input data are created and saved in PeakMatch.

The relationships between experiment and model are summarized in a single, transparent arrangement and made available for assessing the best working conditions.

By creating the chromatographic models from the measured retention data with subsequent special peak assignment methods, continuous peak movements are shown as a function of the eluent properties; Type of organic component, size of the pH value, size of the ionic strength, size of the concentration of additives are determined, the resulting virtual separation system models being compared by comparing the initial experiments with the experiments derived from the computer model and thereby “validated”, ie, they are confirmed in their validity.

The use of a generally known data format (AnDI format), the use of the original measured values, in particular the retention times and the peak areas of the components of a mixture, advantageously allow the components of up to 16 experiments to be assigned.

The object of the present invention is also achieved by a device having the features of claim 33.

The invention is explained in more detail below with reference to the figures using several exemplary embodiments. Show it:

Figure 1: Scheme of the parameter selection of the measurement conditions for automatic model development

Figure 2: Arrangement of four separation runs in four chromatograms each

Figure 3: View of four raw chromatograms and the corresponding peak tables before making the corrections

Figure 4: View of the four chromatograms and peak tables after reducing the displayed values of the peak areas and eliminating irrelevant peaks

FIG. 5: View of an editing of the individual chromatograms for the elimination of the peaks of components which eluted early

Figure 6: View of two chromatograms and peak table with optimized display

Figure 7: View of the four peak tables with marking of the peaks, the arrangement of which is reversed

Figure 8: View of two chromatograms with division of a double peak into the two corresponding single peaks FIG. 9: View of a peak table with the splitting of a double peak into two single peaks when chromatogram 4 is selected as the reference chromatogram

Figure 10: View of a peak table with the splitting of a double peak into two single peaks when chromatogram 2 is selected as the reference chromatogram

Figure 11: View of four matched chromatograms and peak tables

Figure 12: View of a map of the critical resolution and the associated chromatogram after transfer of the data from the matched peak tables

Figure 13: View of options for further processing of the separation model

Figure 14: View of a comparison of simulated and experimentally determined separation runs

Figure 15: View of the results of a method development

Figure 16: View of an overview for the automatic calculation of an optimal gradient

Figure 17: View of a diagram for determining the method robustness R _b

embodiment

The automatic model development according to the invention for HPLC measurements using the software programs ChromMerge, PeakMatch ^® and Drylab ^® is based on a step-by-step optimization of at least two to six experimental parameters. The experimental parameters are, for example, the gradient time tg, temperature T, the pH of the mobile phase, the composition of the mobile phase, the proportion of the organic part or composition of the organic part of the mobile phase, the ionic strength, buffer concentration and any additives such as SDS , Typical organic solvents as part of the mobile phase are acetonitrile ACN, methanol MeOH, ethanol, n- or iso-propanol or tetrahydrofuran THF. The best way to start the optimization is to vary two parameters. The gradient runtime t _G and temperature T are usually selected first. The method can then be further optimized by changing further parameters in at least 2 additional runs

A scheme according to FIG. 1 is used to select the suitable values for temperature T and gradient time t _G in the course of HPLC method development.

The temperature difference between the selected temperature values T1 or T2 is at least 30 ° C, preferably between 30 and 80 ° C. The values for the gradient runtimes tG1 and tG2 should differ by a factor of 3-4. With tG1 they should be 2 or 6 times with tG2 the number of components.

To simplify the subsequent evaluation of the experiments, the same amount should be injected from the same sample in all experiments. Unstable connections should be cooled. Only in this case can one expect the same peak areas for the peak of the same substance in all runs with a standard deviation below 10%.

Experimental chromatograms were recorded on an RP-C18 column, phosphate buffer with a pH of 2.5 as eluent A and 0 to 95% acetonitrile ACN as gradient B. In the first run the gradient run time is tG1 = 30min, at a temperature T1 = 40 ° C, in the second run the gradient runtime is tG2 = 90 min at a temperature T2 = 40 ° C, in the third run tG3 = 30 min at T3 = 70 ° C and in the fourth

Run is tG4 = 90 min and T4 = 70 ° C.

The individual components are detected by their absorption at a certain wavelength or other spectral properties and represented as individual peaks or peak areas in a chromatogram. Instead of the peak areas, molar masses or their bricks or ionized bricks can also be used.

After recording the chromatograms, they are arranged according to FIG. 2. Such an arrangement enables an immediate identification of peak movements in the chromatogram caused by the variable experimental conditions.

The chromatograms are saved in AnDI / AIA format. These chromatograms are imported into the PeakMatch ^® program and can optionally be displayed with a footer with information on the sample or the like (see FIG. 3). Unprocessed chromatograms are often very confusing due to the large peak areas and high number of integrated peaks and hinder easy assignment of the peaks to one another. An important function is the simultaneous numerical representation of the retention times and peak areas in a table arranged below the chromatograms.

The chromatograms shown are simplified in several steps. The peak areas are reduced to small, more legible values, preferably to a maximum of 4-digit numbers. The reduction is done by dividing the values of the peak areas

L0 by a specific factor, which is either automatically specified or can be selected manually. Small peaks that are not of interest can optionally be deleted. A preferred limit value can also be chosen arbitrarily. All peaks with a peak area below this limit are eliminated at two locations: above the respective peaks and in the peak table. After completing these steps, a

Obtained L5 clearer diagram according to Figure 4.

Furthermore, the display of the chromatograms can be further improved using additional options such as zooming, label shift, etc. Advantageously, an option for eliminating peaks that are not displayed or early eluting components is carried out by clicking the “Delete non-zoomed peaks” button, so that the chromatograms are shown in accordance with FIG. 5.

In the following procedure, the retention times and peak area tables are modified in relation to any peak movements in the chromatograms. For this, the

25 standard deviation of the peak areas is used. The values of the standard deviation give an initial overview of the extent to which peaks in the chromatograms have been correctly matched or not. A small standard deviation in a series of similar peak areas is a good indicator that the peaks originate from the same substance and have therefore been correctly adjusted. Such lines get the color green. With a standard deviation of over 10%, however, a clear assignment is no longer easily possible.

It can be seen from FIG. 6 that the first 3 peaks in the peak table have been correctly adjusted; the respective standard deviations are below 10%, so the lines are colored green 5.

The series, which have a standard deviation between 10 and 60%, must be carefully checked. You get a different blue color. The fourth peak in the peak table in FIG. 7 has a standard deviation of 11.2%. Despite something higher standard deviation, this peak is most likely to be correctly adjusted, since the peak in the four chromatograms is very small and the deviating peak areas can be explained by the slightly different position of the baseline and integration. Here it is possible to correct the peak area manually both above the peak and in the table.

A high standard deviation above 60% indicates an incorrect assignment. For example, peaks 5 and 6 in the peak table in FIG. 6 have a standard deviation of 78%. This very high standard deviation is obviously due to a different elution behavior. These lines are shown in red. If there are such red lines, the table can be reused e.g. prohibited for transfer to modeling software.

In the gradient run shorter in time, shown in the left chromatogram in FIG. 6, the small peak 5 with a peak area of 110 elutes earlier than the large peak 6 with a peak area of 873. In contrast, in the time extended gradient run in the right chromatogram in FIG Peak with the large area earlier.

Since the retention data for both the small peak and the large peak must be listed in the same horizontal row, it is necessary to swap the order of the elution data retention time t _r and peak area A in two of the four runs.

If the second run is considered a reference run, the arrangement of the two peaks in the first and third run must be exchanged. For this, the corresponding two peaks from the first run are selected and marked (FIG. 7). After clicking the "Revert Peak elution order" button, the arrangement of the marked peaks (retention time t _R and peak area A) is reversed. If analog peaks are marked in the same two lines in the other runs, all rotation functions can be carried out in one step.

The standard deviation values are updated automatically. After completing the process for the two peaks in the third run, the standard deviations now show values of 2.6% and 1.7%.

In row 7 under peak 7 of the peak table there is another high standard deviation of 28.2

% visible. A comparison of chromatograms 2 and 4 in FIG. 8 clearly shows that this is a double peak. The peak with an area of 59 in chromatogram 2 splits into two peaks with peak areas of 35 and 38 in chromatogram 4. The double peak is selected for correction. When the "Split Double Peak" button is clicked, an information area appears on the screen, which is used to select the number of the run that serves as the reference run. In the example shown, this is run number 4. The area of the selected peak is then divided in the same ratio that the two peaks in chromatogram 4 have (Figure 9).

After dividing the double peak into two single peaks, a high standard deviation shows that the next peaks in the peak table also do not match properly.

The peaks in row 8 are obviously not correctly adjusted (Figure 10). It can be seen from the chromatograms that the peak with an area of 302 in the first chromatogram is a double peak, which is composed of the peaks with an area of 275 and an area of 59 according to chromatogram 2. However, the peak with the area of 59 is also a double peak (see explanations above). In order to share this double peak, the peak is marked according to FIG. 9. By clicking the "Split double peak" button, the information area that asks for the reference run is reactivated. In this case, the reference is chromatogram 2.

The same overlap can be seen in chromatogram 3. The peak with an area of 303 is divided using chromatogram 2 as a reference.

The peak with an area of 366 in chromatogram 4 can also be identified as a double peak. A comparison with chromatogram 2 shows that a

Coelute peak with an area of 275 and a second peak with an area of 120 in chromatogram 4 and coincide. For correction, the corresponding peak with an area of 366 in chromatogram 4 is selected and clicked using the "Split Double Peak" button. Either chromatogram 2 or 3 can be used as the reference chromatogram. However, since the peak with the corresponding area of 120 is obviously not good has been integrated (see baseline and the different areas of the peak in the other runs), chromatogram 3 is used as a reference The peak area 120 can be corrected by double-clicking, whereby a smaller estimated value, for example 90, can be entered in the table there is no need to edit the integration method again to correct the area from 120 to 90.

After identification and correction of the previously identified double peaks, the table according to FIG. 1 1 looks. Starting from the representation of the chromatograms in FIG however, it is also clear that there are three more coelution peaks that need to be corrected. This applies, among other things, to the peak with an area of 410 in chromatogram 4, which is a double peak based on peaks 89 and 331 from chromatogram 2. Click on the peak for correction and use chromatogram 2 as a reference. 5 Corrections are also made for peak 305 in chromatogram 3 and peak 293 in chromatogram 4. The standard deviations for the corrected peaks are still very high. The reason for this is quite obvious: in the short gradient run, the peak with a peak area of 50 to 60 elutes before the larger peak 0 with an area of 340; in the longer gradient run, however, the smaller peak elutes after the large peak. In order to correct the different order of the elution, the corresponding peaks from chromatograms 2 and 4 are marked and the button "Revert Peak Elution Order" is clicked. This leads to the correct arrangement of the peaks. At this point the peak table is correctly adjusted and will be transferred to the DryLab® program Figure 12 shows the result of the peak adjustment and transfer of the data to DryLab®.

> 0 The DryLab® program for creating the virtual separation model for optimal separation is activated by clicking the button with the DryLabΘ symbol in the PeakMatch toolbar and the map of the critical resolution is created.

The representation and evaluation of the model according to FIG. 12 now allows the determination

25 of the optimal values for temperature and runtime of the gradient from the map of the critical resolution. With the map it is possible to display the resolution of the peaks depending on the temperature.

Walking the cursor on the map of the critical resolution enables the display of the 0 separation quality with every possible combination tG-T. The separation conditions are displayed in parallel with the corresponding chromatogram

The separation conditions and separation qualities are marked with different colors on the critical resolution map, whereby the separation conditions are marked in different colors depending on the critical resolution, ie an area marked with red shows optimal separation conditions and an area marked with dark blue shows unfavorable separation conditions or Peak overlaps. The column dimensions and flow rate can be optimized in the "Column Optimization" field. By using different values for column length and diameter, flow rate and particle size, the conditions for high resolution and short analysis time can be determined. With the "Gradient Editor" tab the gradient can be changed manually or the DryLab ^® program calculates the optimal linear gradient for the separation.

Additional optimization options for the model are carried out in the PeakMatch ^® program.

The correct peak match is checked by comparing the original and simulated chromatograms. For this, the option "Comparison" is selected from the "View" menu in PeakMatch® (Figure 13). The program now shows the four experimental input runs together with the chromatograms simulated by DryLab® under the same experimental conditions (Figure 14). If the peak assignment process is correct, the experimental and simulated chromatograms are identical.

The "Rs-Map" option from the "View" menu in PeakMatch® (Figure 13) enables the optimal separation conditions to be determined.

For optimization of gradient vs. Temperature in PeakMatch® the optimal values for temperature and gradient runtime as well as the critical resolution are determined with these settings. To do this, the "Chromatogram" button must be clicked, the map of the critical resolution and the chromatogram corresponding to the entered values are displayed (Figure 15). If other values for temperature and gradient runtime are selected, a chromatogram for the new values is simulated.

The PeakMatch® program enables automatic calculation of a three-step

Gradients for the current separation system. To do this, the program first calculates the best linear gradient. Then two additional gradient points are added and varied systematically over the gradient runtime and over the percentage of component B. The critical resolution is determined for each new gradient. If the critical resolution is greater than a previously determined best value, the new values for the new gradient are stored in a second table.

To carry out the calculation, information according to FIG. 16 must be given. The maximum runtime ("Max time") that is desired for the final gradient, is entered. Furthermore, the increment with which the two additional gradient points are changed along the time axis (“time increment”) and the increment with which the two different gradient points along the axis are changed with the percentage of component B of the eluent (“ Concentration increment "). In both cases, a small value increases the number of simulated gradients, but also increases the time for the necessary calculations. The flow rate of a run can also be specified

The two tables "Current conditions" and "Optimal Tri-segmented gradient" are filled during the gradient optimization with the values for the gradient currently being tested and with the best values of the optimization. Gradient optimization is started by clicking the "Proceed calculations" button. After the optimization process has ended, the gradient with the best critical resolution is stored in the table "Optimal Tri-segmented Gradient" and displayed graphically.

In addition, there is the possibility to test the influence of the column length, the inner diameter and different flow rates to find out whether it is possible to shorten the runtime without impairing the selectivity. For this purpose, an area for the flow rate is specified in the "Flow rate optimization" field and the "Start" button is activated. Analogously, column lengths or inner diameters are entered in corresponding fields to check whether there is room for improvement in the critical resolution.

FIG. 17 shows a diagram for determining the method robustness R _b , the parameters used for the measurement conditions or chromatographic variables such as pH, T _G , gradient% B and buffer _{concentration being} plotted against the values of the critical resolution R _scrιt . The decisive factor for determining the method robustness parameter is the level of the critical resolution a1 and b1 or a2 and b2 at the potential respective working points wp1 and wp2. At the potential working points wp1 and wp2, the critical resolution assumes the values R _s , _wp ι and R _s , w _P 2. The range of fluctuation of the parameters used for the measurement conditions is determined for a specific window w. In the following, the method robustness according to equation (2) is determined from these values. It applies that the value of the robustness is higher the narrower the window of the fluctuation range. The method robustness R _b is for at least 1 up to 8 parameters with R _b = 100 (b / a) (Rs _wp RsRe _f ) with b less than a, the reference value R _{sRef being 1.5} LIST OF REFERENCE NUMBERS

a1, b1, a2, b2 Height of the critical resolution

w _p1 , w _{p2 of} potential operating points

Rs, w _P ι, Rs.w _P 2 Value of the critical resolution at the potential working points w _p1 and w _p2

w Window of the fluctuation range

Claims

claims

1. Process for the automatic model development of chromatographic, electrochromatographic or electrophoretic measurements, in particular in liquid high pressure chromatography (HPLC), capillary electophoresis CE or capillary electrochromatography (CEC),

marked by

a) Separation of at least one substance mixture from at least one sample on at least one column or capillary in individual components and detection of the separated components at the column or capillary outlet after a component-specific retention time t _R on the column or capillary by means of a component-specific signal intensity S in the form of a detector signal, wherein at least one parameter (x, ...), preferably two parameters (x, y, ...), the measurement conditions are changed automatically,

b) Automatic generation of at least one first data set for determining the functional relationship between signal intensity S of the separated components and the retention time t _R in the form of component-specific peaks, the at least first data set taking the at least one parameter (x, ...), preferably two parameters (x, y, ...) to which measurement conditions are assigned,

c) automatic evaluation of the functional relationship of the distance between the component-specific peaks in the form of the critical resolution and the at least one parameter (x, ...), preferably two parameters (x, y, ...), the measurement conditions, and

Generating at least one second data set for creating a virtual separation model based on the determined functional relationship, and

d) generation of at least a third data set for determining a parameter of the method robustness R _b .

2. The method according to claim 1, characterized in that the separation of the substance mixture is carried out at at least two different parameter points of the at least one parameter (x, ...), preferably two parameters (x, y, ...), and at least (2x, ...), preferably [2 (x «y), ...], different separation runs with at least (2x, ...), preferably [2 (xy), ...], different data records generated become.

3. The method according to claim 1 or 2, characterized in that the at least one parameter (x, ...), preferably two parameters (x, y, ...), the measurement conditions from the Group of the gradient time tG, temperature T, pH of the eluent, composition of the eluent from the aqueous and organic phase, composition of the organic phase, ionic strength of the aqueous phase, concentration of additives, in particular SDS, are selected.

4. The method according to at least one of the preceding claims, characterized in that chromatographic separating columns of the type C8 or C18 columns, also with chemically bound ligands which contain polar functional groups, or CE or CEC capillaries with any filling material are used as columns ,

.0

5. The method according to at least one of the preceding claims, characterized in that the component-specific signal intensity S by measuring the light attenuation at a specific wavelength in a wavelength range from 220 to 800 nm of a detector, by measuring the fluorescence, by measuring the

L5 molecular weight (mass spectroscopy) and / or measured by another type of detection.

6. The method according to at least one of the preceding claims, characterized in that after application of the same sample in injection volumes of equal size 0 the value of the specific signal intensity S of a component by the peak area A as a product of the elution volume [ml] and the concentration [μg / ml] of the component is determined, the peak area A being directly proportional to the quantitative mass [μg] of the individual component. 5

7. The method according to at least one of the preceding claims, characterized in that a pictorial representation of the assignment of the at least first data set to the at least first parameter (x, ...) by means of at least (x, ...) diagrams and (x,. ..) tables are made. 0

8. The method according to claim 7, characterized in that a pictorial representation of the assignment of the at least first data set to at least two parameters (x, y, ...) by means of at least [(xy), ...] diagrams and [(xy) , ...] tables.

9. The method according to claim 7 or 8, characterized in that the created 5 diagrams are chromatograms, capillary electrochromatograms or electropherograms.

10. The method according to at least one of claims 7 to 9, characterized in that in the created at least 2 chromatograms, the displayed values of Peak areas A, which are plotted over the peaks, are simultaneously reduced by a factor calculated by the software and / or peak areas of small peaks and / or peaks of components which elute early are eliminated.

11. The method according to at least one of claims 7 to 10, characterized in that the displayed values of the peak areas in the at least (x, ...) peak area tables are simultaneously reduced by a factor calculated by the software.

12. The method according to at least one of claims 7 to 1 1, characterized in that each peak defining a single component is localized in the at least (x, ...) diagrams, preferably [(xy)], ...] diagrams a data set is generated from the functional relationship between retention time t _R and peak area A for each peak and is summarized in at least (x, ...), preferably [(xy), ...], peak tables.

.5 13. The method according to claim 12, characterized in that the localization of a peak defining a single component in the at least (x, ...), preferably [(xy)], ...], diagrams by considering the standard deviation of the corresponding Peak areas A occur in the at least (x, ...), preferably [(xy)], ...], peak tables, with a standard deviation of less than or equal to 10%, a correct assignment of the peaks defining an i θ individual component in the at least (x , ...), preferably [(x * y)], ...], means diagrams and a standard deviation of greater than or equal to 60% means an incorrect assignment of the peaks.

14. The method according to claim 12 or 13, characterized in that a correct assignment of the peaks with a standard deviation over 10% is carried out manually in the peak table.

15. The method according to at least one of the preceding claims, characterized in that the virtual separation model comprises a data record with unambiguous 0 data assignment of the critical resolution to the corresponding parameters of the signal intensities S and retention times t _{R of} a virtual separation run.

16. The method according to claim 15, characterized in that the critical resolution is determined by the at least one parameter (x, ...), preferably two parameters (x, y, ...) of the 5 measurement conditions,

17. The method according to claim 15 or 16, characterized in that for each arbitrarily selected value of the at least one parameter (x, ...), preferably two parameters (x, y, ...), a virtual separation run from the data set critical resolution can be determined.

18. The method according to at least one of the preceding claims, characterized in that information about column dimensions, column filling material, grain diameter or flow rate can be changed virtually.

19. The method according to at least one of the preceding claims, characterized in that information about the gradient of the eluent can be changed.

20. The method according to claim 20, characterized in that the gradient of the eluent can be changed automatically with at least 3 or more segments.

21. The method according to at least one of the preceding claims, characterized in that a comparison of the experimental and virtual separation runs is carried out to control the model validation.

22. The method according to at least one of the preceding claims, characterized in that an automatic three-stage gradient for a separating run can be modeled.

23. The method according to claim 22, characterized in that an optimal linear gradient by inserting at least two additional gradient points specifying the maximum runtime (max time), specifying the step size with which the two different gradient points are changed along the time axis (time increment) Specification of the step size with which the two different gradient points along the axis are changed with the percentage specification of component B of the solvent (concentration increment), and specification of the flow rate of a run (flow) can be simulated.

24. The method according to claim 22 or 23, characterized in that a) maximizes the critical resolution, b) searched for the shortest analysis time, and c) after specifying the

Flow rate (Flow) the shortening of the analysis time is modeled by calculating the changed selectivity.

25. The method according to at least one of the preceding claims, characterized in that the method robustness R _b for at least 1 to 8 parameters

R _b = 100 (b / a) (R _Sw p / RsRef) (2) is determined, where a and b are the height of the critical resolution with b less than a, and R _{Swp is} the critical resolution at the potential operating point and R _{SRef is} the reference value of the critical resolution.

26. The method according to claim 25, characterized in that the permitted fluctuation ranges of the parameters are determined individually by suitable algorithms per parameter.

27. The method according to at least one of the preceding claims, characterized in that the time axis sections are calculated and displayed graphically at the same time for all other chromatograms after a reference run in accordance with the image section created there.

28. The method according to at least one of the preceding claims, characterized in that the following working methods are set up:

-% B with 2-16 runs - tG with 2-16 runs

- pH, concentration of additives, with 3-16 runs each

- tG vs temperature (with 4-16 runs) - tG vs pH (with 6-16 runs)

- tG vs ternary eluent composition (with 6-16 runs)

29. The method according to at least one of the preceding claims, characterized in that it is no longer necessary to save the DryLab models as ^* .inp files, since the input data are created and stored in PeakMatch.

30. The method according to at least one of the preceding claims, characterized in that the relationships between the experiment and the model are combined in a single, transparent arrangement and made available for assessing the best working conditions.

31. The method according to at least one of the preceding claims, characterized in that the creation of the chromatographic models from measured retention data with subsequent special peak assignment methods, which represents continuous peak movements as a function of the eluent properties: type of organic component, size of the pH value, size the ionic strength, the size of the concentration of additives, the resulting virtual separation system Models are compared by comparing the initial experiments with the experiments derived from the computer model and thereby "validated", ie their validity is confirmed.

32. The method according to at least one of the preceding claims, characterized in that by using a generally known data format (AnDI format), by using the original measured values, in particular the retention times and the peak areas of the components of a mixture, an assignment of the components of up to 16 experiments possible.

33. Device for automatic model development of chromatographic, electrochromatographic and electrophoretic methods according to a method according to at least one of the preceding claims.