PROCESSING OF SPATIALLY RESOLVED, ION-SPECTROMETRIC MEASUREMENT SIGNAL DATA TO DETERMINE MOLECULAR CONTENT SCORES IN TWO-DIMENSIONAL SAMPLES

Abstract

The invention relates to methods for processing ion-spectrometric measurement signal data which are recorded spatially resolved across a two-dimensional sample, comprising: —providing the measurement signal data which have a plurality of measurement signal histograms, where a histogram contains a measurement signal tuple with intensity dimension (J), mass dimension (m), and collision cross-section dimension (σ), or quantities derived therefrom; —specifying first and second selections of ionic species for the sample, whose presence in histograms is detectable and distinguishable using the collision cross-section dimension or proxy; —determining the spatially resolved content of ionic species from the first and second selections in histograms of the finite areas (A.sub.fin,x,y), and computing the various contents to form spatially resolved content scores (G.sub.x,y); and —labeling the sample with the content scores (G.sub.x,y). The invention also relates to methods for acquiring and processing ion-spectrometric measurement signal data, and ion mobility-mass spectrometers.

Claims

1. A method for processing ion-spectrometric measurement signal data which are recorded spatially resolved across a two-dimensional sample, comprising: providing the measurement signal data, which have a plurality of measurement signal histograms, where a measurement signal histogram is assigned, by means of two location coordinates (x, y), to a finite area (A.sub.fin,x,y) of the two-dimensional sample, which is smaller than a total area (A.sub.total) of the two-dimensional sample, and contains a measurement signal tuple having intensity dimension (J) or a quantity derived therefrom, mass dimension (m) or a quantity derived therefrom, and collision cross-section dimension (σ) or a quantity derived therefrom; specifying a first selection of ionic species and a second selection of ionic species for the two-dimensional sample, whose presence in measurement signal histograms is detectable and can be distinguished using the collision cross-section dimension (σ) or the quantity derived therefrom: determining the spatially resolved content of ionic species from the first selection and the spatially resolved content of ionic species from the second selection in measurement signal histograms of the finite areas (A.sub.fin,x,y), and computing the various contents to form spatially resolved content scores (G.sub.x,y); and labeling the two-dimensional sample with the spatially resolved content scores (G.sub.x,y).

2. The method according to claim 1, wherein the two-dimensional sample was prepared with a matrix substance for matrix-assisted laser desorption.

3. The method according to claim 1, wherein the first selection of ionic species comprises those of high analytical interest, and the second selection of ionic species comprises those of low analytical interest.

4. The method according to claim 1, wherein the first selection of ionic species comprises biomolecules such as proteins, peptides, glycans, and/or lipids in the two-dimensional sample.

5. The method according to claim 1, wherein the second selection of ionic species comprises charged atoms or molecules and/or clusters thereof, which are generated by the method of sample preparation and/or the method of ionization.

6. The method according to claim 1, wherein after the labeling, a user is presented with an image of the two-dimensional sample in which individual finite areas (A.sub.fin,x,y) are visibly coded with the assigned content scores (G.sub.x,y).

7. The method according to claim 1, wherein predetermined value ranges of the content score are evaluated as ion-spectrometric measurement signals from outside the two-dimensional sample.

8. The method according to claim 1, wherein a subsequent evaluation (i) only takes into account measurement signal histograms from finite areas (A.sub.fin,x,y), where the content scores (G.sub.x,y) lie in a predetermined range of values, and/or (ii) uses the content scores (G.sub.x,y) as weighting factors.

9. The method according to claim 1, wherein the first selection of ionic species is specified by summing several measurement signal histograms into an aggregated measurement signal histogram and determining an interesting portion of measurement signal tuple entries in the aggregated measurement signal histogram.

10. The method according to claim 9, wherein the interesting portion of the measurement signal tuple entries is used to distinguish the first selection of ionic species from the second selection in at least one dimension of the measurement signal histograms.

11. The method according to claim 9, wherein the interesting portion of measurement signal tuple entries is determined by means of regression analysis.

12. The method according to claim 11, wherein the regression analysis comprises a logarithmic regression or a logarithmic Radon transform.

13. The method according to claim 11, wherein the regression analysis searches for a correlation between collision cross-section (σ) and mass (m) of a molecule according to the equation σ(m)≈C m.sup.k, where C is a molecule-dependent proportionality factor and k is a molecule-dependent exponent.

14. The method according to claim 9, wherein the aggregated measurement signal histogram is calculated by a location-independent summation of several measurement signal histograms.

15. The method according to claim 14, wherein the location-independent summation takes into consideration only those measurement signal histograms where the measurement signal tuple entries of at least one dimension (i) exceed a predetermined threshold value, (ii) are below a predetermined threshold value, or (iii) are within a predetermined value range.

16. The method according to claim 14, wherein at least one measurement signal tuple entry of the individual measurement signal histograms is transformed before the summation such that measurement signal tuple entries of a first predetermined range are disproportionately weighted with respect to a second predetermined range.

17. The method according to claim 1, wherein the spatially resolved content score (G.sub.x,y) is calculated using $G_{x,y}=\frac{{\Sigma }_{i\in S_a}{J}_i}{{\Sigma }_{i\in S_a}{J}_i+{\Sigma }_{{\mathrm{iϵS}}_b}{J}_i}\mathrm{or}{G}_{x,y}=\frac{{\Sigma }_{i\in S_a}{J}_i}{{\Sigma }_{i\in S_b}{J}_i},$ where S.sub.a and S.sub.b respectively designate the quantity of those indices i for which the corresponding measurement signal tuple entries (m.sub.i, σ.sub.i) of the individual measurement signal histogram were assigned to the first and second selection of ionic species, respectively.

18. A method for acquiring and processing ion-spectrometric measurement signal data, including: (i) acquiring the ion-spectrometric measurement signal data using an ion mobility spectrometer-mass spectrometer, and (ii) executing a processing method according to claim 1.

19. An ion-mobility spectrometer-mass spectrometer having a computing and/or control unit which is designed and configured to execute a method according to claim 1.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The invention can be better understood by referring to the following illustrations. The elements in the illustrations are not necessarily to scale, but are primarily intended to illustrate the principles of the invention (mostly schematically).

[0026] FIG. 1 depicts a collision cross-section-mass signal histogram with highlighted trend lines for classes of molecule from the Prior Art paper of Shelley N. Jackson et al. (J. Mass Spectrom. 2007 August; 42(8); 1093-1098; FIG. 1 with caption there).

[0027] FIG. 2 depicts an adapted schematic representation of an ion mobility spectrometer-mass spectrometer from the Prior Art paper by Jeffrey Spraggins et al. (Anal. Chem. 2019, 91, 22, 14552-14560; FIG. 1 there), with which collision cross-section-mass signal histograms (e.g., location coordinates x, y; signal tuple m, σ, J), can be acquired. The special method of mobility separation (parallel accumulation trapped ion mobility spectrometry—PATIMS) is described in more detail in the published patent applications EP 3 054 473 A1 or US 2016/0231275 A1 of the applicant, for example.

[0028] FIG. 3 shows the schematic representation of an example of a flat sample in the form of a tissue section for illustrative purposes, which here imitates the thin section of a mouse brain, once with (bottom left) and once without (top right) square grid for the points or pixels of an ion-spectrometric image.

[0029] FIG. 4 shows an example of a collision cross-section-mass signal histogram of a MALDI IMS-MSI measurement. The mass (or more precisely the mass-to-charge ratio, designated by m/z) is plotted on the horizontal axis; the collision cross-section (expressed by the derived quantity 1/K.sub.0) is plotted on the vertical axis.

[0030] FIG. 5 is a logarithmic depiction of an aggregated collision cross-section-mass signal histogram which shows the regression line (solid white line) and a signal corridor (dashed line) for the dominant signal component, which is of interest.

[0031] FIG. 6 depicts the Radon transform of the logarithmic collision cross-section-mass signal histogram from FIG. 5. The projection angle is plotted on the horizontal axis, the distance from the center of the logarithmic histogram is plotted on the vertical axis. The white cross identifies the position with maximum intensity.

[0032] FIG. 7 shows the spatial distribution of a content score for a MALDI IMS-MSI lipid measurement. In the control area outside of the tissue sample (square field bottom left), the scores determined are significantly lower than in the tissue itself. The dark regions within the sample correspond to the denser tissue zones with a higher portion of lipid signals relative to the matrix background.

DETAILED DESCRIPTION

[0033] While the invention has been illustrated and explained with reference to a number of embodiments, those skilled in the art will recognize that various changes in form and detail can be made without departing from the scope of the technical teaching, as defined in the attached claims.

[0034] Hereinafter, a method for assessing spatially resolved ion mobility spectrometry-mass spectrometry data is described which provides a score or quantifier for every area indicating approximately what percentage of the spectra measured there originates from species of analytical interest from the sample. A high value for this quantifier means a high proportion of molecules of the species of interest ionized from the sample and thus a high information content and a high signal quality. Low values, in contrast, can indicate that the measured spectra are dominated by a background signal, which can be traced back to the method of sample preparation and/or the method of ionization, for example.

[0035] FIG. 2 is a schematic representation of a possible ion mobility spectrometer-mass spectrometer with which collision cross-section-mass signal histograms from a two-dimensional or flat tissue sample can be acquired with spatial resolution, see setup in Part A. The setup and operation shall be explained here very briefly:

[0036] A laser system (Bruker SmartBeam 3-D, top left) with various optical components is designed to bombard a flat sample on a sample support with pulses. The sample support can be scanned step-by-step to obtain spatially resolved measurement signals from the flat sample, e.g., a wide two-dimensional sample such as a flat tissue section or an array of separately prepared samples such as locally applied tissue homogenate preparations.

[0037] Once generated, the ions enter the ion mobility spectrometer (dual TIMS cell), which has an accumulating section and an analyzing section. An inert gas flows through both sections of the dual TIMS cell (from left to right in the illustration). In the cell, the gas flow drives the ions against an opposing electric field, see Detail B in the center for the illustrated principle. In the analyzing section, the ions are separated according to their mobility at different positions along the axis.

[0038] An incremental decrease in the electric field strength in the analyzing section of the dual TIMS cell allows a sequential release of the ions separated according to their mobility (Detail B, Scan). After the mobility analysis in the analyzing section, the ions which have meanwhile collected in the accumulating section are transferred to the analyzing section (Detail B, Pulse). The ions exiting the analyzing section initially pass through an ion transfer multipole and then enter a quadrupole mass filter. Here, ions can be selected for further analysis, while other ions can be removed. The ions are subsequently transferred into a collision cell, where the ions selected are fragmented by accelerated injection into a neutral gas.

[0039] The ions and/or any fragment ions produced therefrom are stored temporarily in the collision cell, before being introduced into the ion pulser of a time-of-flight analyzer with orthogonal injection, temporally coordinated. There they are accelerated perpendicular to the direction of injection onto the flight path of the reflector time-of-flight analyzer. At the end of the flight path, a detector (not shown) receives the different ion packets with temporal and hence mass resolution, and outputs them as a time-of-flight transient, which can subsequently be rescaled into masses (m) or mass/charge ratios (m z).

[0040] Above, a so-called trapped ion mobility spectrometer (TIMS) coupled to a time-of-flight mass analyzer was described with reference to FIG. 2. It shall be understood that other embodiments of mobility spectrometers and mass analyzers can be used for the purpose of generating collision cross-section-mass signal histograms, for example a drift tube ion mobility spectrometer, in which the neutral gas is at rest or flows in the opposite direction to the ion motion, and an electric drawing field is used and which is coupled with a cyclotron resonance cell, for example.

Description of an Exemplary Method

[0041] The method described here is suitable to be applied to IMS data acquired in connection with a MALDI MSI ion source (MALDI IMS-MSI). In a preferred embodiment, the method comprises at least the following steps:

1. Calculating an aggregated collision cross-section-mass signal histogram which collates the information on the individual measuring points of the flat sample from several (preferably all) histograms.
2. Determining a signal portion of interest (first selection of ionic species), for example by means of regression analysis to determine the correlation between collision cross-section and mass for the dominant signal portion of the complete measurement.
3. Specifying a section of the collision cross-section-mass plane (σ-m plane or a plane of quantities derived therefrom), which contains the dominant and interesting signal portion of the measurement (“signal corridor”), if applicable accompanied by specification of a second section of the collision cross-section-mass plane with a further signal portion which is of lesser interest, for example.
4. Calculating a content score, which can be termed a signal quality score, for individual measuring points of the two-dimensional or flat sample by comparing the signal portions within and outside of the signal corridor or within the two previously defined (preferably disjunct) sections (e.g., covering different charge states (z) of molecules and/or classes of molecules).
5. Evaluating of the content score for the subsequent analytical steps. Hereinafter, these steps are described in more detail.

Aggregation of the Collision Cross-Section-Mass Signal Histograms

[0042] In the first step, an aggregated collision cross-section-mass signal histogram which collates the information on the individual measuring points from preferably all histograms can be formed. To shorten the computing time, it is also possible to use a representative selection of the histograms to form the aggregated collision cross-section-mass signal histogram, where necessary. The aggregation can, for example, be carried out by simple sum formation, i.e., the individual histograms are summed in terms of intensity (J) to form a sum histogram. More complex aggregations are also conceivable, however, e.g., such that in each histogram that is included, only dominant signal portions above a specific threshold value of the intensity (J.sub.min) are considered. The signal intensities of the individual histograms could also be modified by a transform before the sum is formed, e.g., to place greater or less emphasis on the portions with a strong signal compared to those with a weaker signal, depending on the objective of the evaluation.

Regression Analysis

[0043] It is assumed that the correlation between collision cross-section a and mass m for the signal portion of interest (for example the dominant one) of the complete measurement (or the correlation for correspondingly derived quantities) is approximately described by a power law with parameters C and k, as explained above. These parameters can be estimated on the basis of the aggregated histogram by using a suitable regression analysis. The aggregated histogram is composed of a number of individual signals, with each individual signal corresponding to a tuple (e.g., m.sub.i, σ.sub.i, J.sub.i), where m.sub.i designates the respective mass (possibly also (m/z).sub.i, mass-to-charge ratio), σ.sub.i the collision cross-section (possibly also a derived mobility quantity K.sub.i) and J.sub.i the measured signal intensity. Several methods are possible for the actual regression analysis, including the following in particular:

Simple Logarithmic Regression

[0044] By taking the logarithm of the power law shown above, a linear relationship is obtained so that the measurements are described by the model equation

log(σ.sub.i)=k log(m.sub.i)+log(C)+ε.sub.i.

Here ε.sub.i designates the deviations from the model, which are assumed to be randomly distributed. By means of simple linear regression, the constants C and k can be determined via

[00004]$k=\frac{M_{00}{M}_{11}-M_{10}{M}_{01}}{M_{00}{M}_{20}-{\left(M_{10}\right)}^2}$$\mathrm{and}$$\log \left(C\right)=\frac{1}{M_{00}}\left(M_{01}-k{M}_{10}\right)$$\mathrm{with}$$M_{\mu v}=\underset{i}{.Math.}{J_i\left(\log m_i\right)}^{\mu }{\left(\log {\sigma }_i\right)}^v,\mu ,v\in \left\{0,1,2\right\}$

The above equations can easily be adapted for quantities derived from the collision cross-section (σ) and mass (m).

Robust Logarithmic Regression

[0045] A known disadvantage of the simple logarithmic regression described above is the sensitivity to outlier values in the measured data. Alternatively, more robust regression methods can be used to estimate the parameters k and log(C) of the linear model equation. Two possible methods in particular can be used, namely (i) the Lasso regression (Tibshirani, Robert (1996). “Regression Shrinkage and Selection via the lasso”. Journal of the Royal Statistical Society. Series B (methodological). Wiley. 58 (1): 267-88) and/or (ii) the Theil-Sen estimator (see Sen, Pranab Kumar (1968), “Estimates of the regression coefficient based on Kendall's tau”, Journal of the American Statistical Association, 63 (324): 1379-1389).

Logarithmic Radon Transform

[0046] A further possibility for the regression analysis is to apply the Radon transform to a 2D histogram of the measured intensities in the logarithmic σ-m plane (see Radon, Johann (1917), “Über die Bestimmung von Funktionen durch ihre Integralwerte langs gewisser Mannigfaltigkeiten”, Berichte Uber die Verhandlungen der Königlich-Sachsischen Akademie der Wissenschaften zu Leipzig, Mathematisch-Physische Klasse, Leipzig: Teubner (69): 262-277). To this end, the range of values covered by the logarithmic measurement values (log m.sub.i, log σ.sub.i) is subdivided into rectangular subsections of equal size. For each subsection, those intensities J whose associated measured values (log m.sub.i, log σ.sub.i) fall into this subsection are summed (see FIG. 5). The Radon transform is applied to the resulting 2D histogram and the position with maximum intensity is sought within it (see white cross in FIG. 6). The inverse transform of this position gives the required straight line in the logarithmic σ-m plane (see FIG. 5). The parameters of this straight line—the gradient k and the y-intercept log(C)—finally describe the required correlation between collision cross-section σ and mass m according to the power law in the form of a curve which represents the profile of the dominant and interesting signal portion in the aggregated collision cross-section-mass signal histogram. Here also, the method can be adapted to quantities derived from the collision cross-section (σ) and mass (m).

Specifying a Corridor or Region of Signal Portions

[0047] After the proportionality factor C and the exponent k of the power law have been determined, factors C.sub.lo<C and C.sub.hi>C can be set (lo stands for low; hi for high). With these values in place of C in the power law, the result is shifted curves, which specify the boundaries of the signal corridor (see FIG. 5). By suitably selecting C.sub.lo and C.sub.hi, these curves enclose the dominant signal portion. By selecting C.sub.lo=0, the signal corridor can be extended downward and to the right as far as the boundaries of the histogram so that the corridor is actually only bounded at the top. Alternatively, the signal corridor can be extended to the top left by C.sub.hi=∞ (infinite). What constitutes a suitable selection of C.sub.lo and C.sub.hi depends on the class of molecule under analysis. Typical values for lipids and glycans are C.sub.lo=0.90 C . . . 0.96 C and C.sub.hi=1.04 C . . . 1.10 C, for example. In an alternative embodiment, a similar method can be used to determine a second section of the collision cross-section-mass plane (second selection), which should not contain all the measurement signals outside the previously determined signal corridor or range but only specific ionic species, especially those of species of no interest, such as MALDI matrix clusters or similar background. Here also, the method can be adapted to quantities derived from the collision cross-section (σ) and mass (m).

Calculation of the Content Scores (“Signal Quality Scores”)

[0048] The content scores for each individual measuring point are computed with the aid of the corresponding collision cross-section-mass signal histogram. The relevant tuple entries for an individual measuring point or an individual histogram are designated, as above, by (m.sub.i, σ.sub.i, J.sub.i), for example. S designates the quantity of those indices i, for which the corresponding measurement values (m.sub.i, σ.sub.i) lie within the signal corridor or range. The content score G can then be calculated for the histogram under consideration using

[00005]$G=\frac{{\Sigma }_{i\in S}{J}_i}{{\Sigma }_{i\in S}{J}_i+{\Sigma }_{i.Math.S}{J}_i}.$

[0049] In this example, the measurement signals from the signal corridor or range are therefore ratioed with all measurement signals (lying within and outside of the corridor or range). This method of computation has proved to be particularly stable in data processing terms. The second selection of ionic species (e.g., the signal portion of no interest) comprises all the measurement signals that do not lie in the preferred signal corridor or range, which in this embodiment in turn determines the measurement signals of the ionic species of the first selection.

[0050] In further embodiments, however, a method of computation is also possible whereby the first and second selection of ionic species are ratioed directly e.g., as per the following equation:

[00006]$G_{x,y}=\frac{{\Sigma }_{i\in S}{J}_i}{{\Sigma }_{i.Math.S}{J}_i}.$

This computation method provides a content score that represents the actual content ratios of each individually selected ionic species and is therefore easier to interpret.

[0051] It is, furthermore, conceivable and intended to define the second selection of ionic species such that it does not include all the ionic species that were not assigned to the first selection of ionic species. For example, a second signal corridor or range can be determined for a specific second signal portion in a collision cross-section-mass plane (or in a plane of correspondingly derived quantities). This approach can be useful when a measurement signal histogram contains signal portions of several different substance classes, e.g., lipids, glycans, and peptides, which as biomolecules are in principle candidates for a signal portion of interest, as well as matrix clusters or other background ion species which are essentially never of interest. The content score can be computed as before with:

[00007]$G_{x,y}=\frac{{\Sigma }_{i\in S_a}{J}_i}{{\Sigma }_{i\in S_a}{J}_i+{\Sigma }_{{\mathrm{iϵS}}_b}{J}_i}\mathrm{or}{G}_{x,y}=\frac{{\Sigma }_{i\in S_a}{J}_i}{{\Sigma }_{i\in S_b}{J}_i},$

where S.sub.a and S.sub.b respectively designate the quantity of those indices i for which the corresponding measurement signal tuple entries (e.g., m.sub.i, σ.sub.i or quantities derived therefrom) of the individual measurement signal histogram were assigned to the first or second selection of ionic species respectively. Furthermore, the second computation method above has the advantage that the content score can easily be inverted, thereby producing an informative result, for example when the first and second selections of ionic species comprise different biomolecules from the two-dimensional or flat sample and ratio them to each other.

Evaluation of the Content Scores (“Signal Quality Scores”)

[0052] The content scores can be evaluated by visualizing the spatial distribution of the spatially resolved content scores as a gray scale or false-color image (see FIG. 7). As an alternative, the content scores can be used to control a subsequent analysis of the data. For example, it is possible to use only those measuring points and histograms for the analysis whose content scores are above a specific threshold value, below a specific maximum value, or within a specific value range. The content score could also be used as a weighting factor so that those measurement data with a higher content score (and thus presumably of better quality) receive a higher weighting than those with a low score. The different weightings can be used for adaptive noise suppression in a subsequent data analysis, for example.

[0053] Further embodiments of the invention are conceivable in addition to the embodiments explained by way of example. With knowledge of this disclosure, those skilled in the art can easily design further advantageous embodiments, which are to be covered by the scope of protection of the claims, including any equivalents as the case may be.