Direct and automated chromatography-mass spectrometry

Abstract

A method, spectral detection system, and computer-readable medium for collecting spectral data for a sample, comprising: detecting the presence of a compound within a time window; performing a spectral library search using the spectral data from the time window; evaluating a hit list of compounds within each time window and selecting a subset of likely compounds; performing a regression analysis between retention index and measured retention time for the compounds in the subset; identifying and removing outliers from the subset with large retention index error; repeating the regression analysis where the outliers are removed; calculating retention index values ​​for all compounds from the entire sample run; comparing the calculated retention index values ​​with retention index values ​​of possible compounds to be identified; and using the retention index match score as an additional metric or filter to further evaluate the likelihood of possible hits.

Description

[Technical field]

[0001] The present invention relates generally to the field of chromatographic separations coupled to spectral detection systems such as gas chromatography (GC) with mass spectrometry (MS) detection, and more particularly to methods for acquiring, processing, and analyzing the resulting separations and spectral data. [Background technology]

[0002] Ervin Kovats (Kovats E., Helv. Chim. Acta 41, 1915-1932 (1958)) introduced the use of a chromatographic retention time index called the Kovats Retention Index (KRI or simply RI), a dimensionless number sometimes expressed as Index Units or simply "iu", as a method of converting chromatographic retention times (RT) to system-independent constants in gas chromatography (GC). The method relies on the calibration of the chromatograph, usually with a series of n-alkanes, from which the Kovats index of an n-alkane is defined as 100 times the number of carbon atoms in the n-alkane, e.g., n-hexane (C 6 H 14 ) is assigned an RI value of 600iu. Organic compounds can then be assigned RI values ​​relative to these standards. Models have been developed for both isothermal GC (logarithmic function) runs as well as temperature programmed runs (linear interpolation function).

[0003] RI can be a powerful tool to aid in unknown organic compound identification, provided that the RI of the unknown compound is available. The widely used NIST / EPA / NIH Mass Spectral Library updates and maintains a compound database of measured RI values ​​for nearly 140,000 compounds (https: / / www.nist.gov / programs-projects / nist20-updates-nist-tandem-and-electron-ionization-spectral-libraries). In addition, NIST is developing artificial intelligence (AI) models (see, e.g., Matyushin, DD et al., Int. J. Mol. Sci. 22(17), 9194 (2021) or Stein, SE et al., J. Chem. Inf. Model., 47(3), 975-980 (2007)) that can calculate RI from chemical structures with high accuracy and provide RI values ​​for nearly all of the over 300,000 compounds in the electron ionization mass spectrometry (EI-MS) database.

[0004] Although RI values ​​alone do not necessarily uniquely identify an unknown compound, when coupled to spectral library searching in EI-MS databases (e.g., Stein, SE et al., J. Amer. Soc. Mass Spectrom. 5, 859-866 (1994)), a much higher degree of confidence in unknown identification can be achieved than by either technique alone. For example, library searching typically produces a list of the "best" possible matches of the unknown to the library spectrum. However, this list does not usually provide a definitive or unique identification. Similarly, RI does not provide a definitive identification, since many compounds can have similar RI values. However, retrieving both dimensions of information simultaneously can significantly improve identification confidence.

[0005] To accurately calculate the RI values ​​of unknowns, a calibration of known compounds, usually an n-alkane ladder, must be carefully performed under the same conditions as those of the sample run, including flow rates, temperature program, inlet temperature, column, etc. Since the RI values ​​are calculated by interpolating between the RI standards, it is optimal to have these standards evenly spaced and close to each other across the chromatographic run. Any change in method or conditions (flow rates, column length, etc.) requires the calibration to be re-run. In addition, over time, chromatographic conditions may change slightly due to ambient temperature, column aging, or simply small drifts in the GC electronics and flow controllers, which will require re-calibration. And of course, running a new chromatographic method requires the RI calibration to be re-run. Thus, the disadvantage of this approach is that the additional time and effort required to routinely re-run the calibration samples imposes an additional burden and leads to reduced analytical throughput and efficiency. Another difficulty in using external standards such as n-alkanes is that some alkanes may not elute from the column before the GC programming is complete, leading to carryover or other experimental complications, or the injected alkane standard mixture may not contain alkanes of high enough alkane number to cover the full retention time range of interest.

[0006] Internal calibration (introducing a known compound into the run) can also be used to calibrate the GC. This is usually accomplished by adding a known compound into the sample run with a known RV value. The advantages of this approach are: 1) there is minimal time lapse between the calibration standard and the analyte, which minimizes errors due to instrument drift; 2) the internal standard ensures that the calibration standard and the sample are run under identical conditions to maximize accuracy; and 3) the approach eliminates the need for a separate calibration run, thus saving time and effort. However, many samples are very complex as is, and adding a standard into the sample can further complicate the analysis with peak coelution, ion suppression, etc., making it difficult to accurately determine the retention time (RT) of the standard. Finally, this approach also requires an additional step to ensure that the amount of RI standard added into the sample corresponds to the amount of analyte contained in the sample itself.

[0007] Finally, RI models developed for isothermal (logarithmic model) and temperature programmed runs (linear interpolation model) can introduce significant errors when applied to runs that combine multiple temperature ramps or ramps combined with isothermal segments (see example in Figure 1). Unfortunately, multi-ramp temperature programs are the norm in GC due to the ability of these programs to maximize separation power and minimize run time. Although advanced RI models have been proposed to accurately calculate RI values ​​within such runs, advanced RI models require precise measurements of three thermodynamic constants ΔH, ΔS, and ΔCp for each compound as well as careful characterization of the gas chromatograph (Boswell, P.G. et al., J. Chromatogr. A. 1263:179-88 (2012) and Peng.B. et al., J. Chromatogr. A. 1374:207-215 (2014)). Unfortunately, unlike traditional KRIs, these thermodynamic constants are not available for the vast majority of organic compounds. To measure these constants accurately, one would need to run samples containing these compounds under at least three clearly different temperature programming conditions under which all these compounds would elute, a difficult experimental setup and a challenging task, and therefore impractical. [Prior art documents] [Non-patent literature]

[0008] [Non-Patent Document 1] Kovats E., Helv.Chim.Acta41, pp. 1915-1932 (1958) [Non-Patent Document 2] Matyushin, DD et al., Int.J.Mol.Sci.22(17), 9194 (2021) [Non-Patent Document 3] Stein, S. E. et al., J. Chem. Inf. Model., 47(3), pp. 975-980 (2007) [Non-Patent Document 4] Stein, S. E. et al., J. Amer. Soc. Mass Spectrom. 5, 859-866 (1994). [Non-Patent Document 5] Boswell, PG et al., J. Chromatogr. A. 1263: 179-88 (2012) [Non-Patent Document 6] Peng, B. et al., J. Chromatogr. A. 1374:207-215 (2014) Summary of the Invention [Problem to be solved by the invention]

[0009] Therefore, it would be desirable and highly advantageous to have a method that overcomes the above-noted deficiencies and shortcomings of the prior art. [Means for solving the problem]

[0010] The present application is directed to the following improvements: 1. A new method of RI calibration without the use of any known standards beforehand, either added to the test sample itself or by external sample injection or run, that saves time and effort while avoiding experimental, laboratory, or human error. 2. These RI standards are self-generated and identified by the sample analysis itself, using a subset of analytes from the very same samples. 3. A technique for spectrally selecting an analyte from a test sample run to be used as the RI standard for calibration of an entire sample run under the same experimental or temperature programming conditions between the standard and the analyte from the test sample. 4. A technique for iteratively refining or improving the set of standards used in RI calibration. 5. RI calibration is internal to the run, so the RI calibration is independent of the separation conditions. 6. Application of calibrated (approximated) or calculated RI values ​​to each detected resolved peak to further aid its identification, either in the form of additional metrics beyond the spectral library search or in combination with the spectral library search scores, which can then be used to color code any identified compound in the list of compounds on a computer screen to indicate the likelihood of a correct identification, greatly reducing the burden on the analyst performing the analysis. Such additional identification quality metrics provide greater confidence than either the spectral identification or the RI value identification used alone. 7. Allows retrospective calculation of RI values ​​from previously performed runs that did not have conventional RI calibration to improve confidence in compound identification. 8. The RI values ​​thereby obtained can be fed back into an existing spectral library or database to complement and enhance RI collection to continually increase spectral discrimination capabilities. 9. Prior art has shown that it is also possible to calibrate quadrupole MS systems for both exact mass and spectral accuracy, and therefore determine the elemental composition of unknown compounds not yet included in a spectral library or database, in which case the RI calculated from its RT can be compared to the RI predicted from a computer model based on possible chemical structures of the determined elemental composition in order to determine, select, rank, or project likely chemical structures.

[0011] Each of these aspects is described below along with experimental results to demonstrate their utility. [Brief description of the drawings]

[0012] [Figure 1] FIG. 1 is an example of a temperature program for GC-MS analysis. [Diagram 2] FIG. 1 is a block diagram of a mass spectrometer system coupled to a separation device that can utilize the methods disclosed herein. [Diagram 3]FIG. 2 is a total ion chromatogram (TIC) from a GC / MS analysis of a thermally aged fragrance sample. [Figure 4] FIG. 1 is a diagram of the temperature program for GC-MS analysis of thermally aged fragrance samples. [Diagram 5] Plot of RI values ​​from the NIST database from the top spectral search hits for each peak against RT for thermally degraded fragrances. It is also visually apparent that there is a good correlation, but that there are many outliers due to compounds that were missed when using only the spectral search. [Figure 6] Plot of RI values ​​from the NIST database from the top spectral search hits for each peak against RT for thermally degraded fragrances after discarding all compounds that do not meet the spectral search quality criteria. The final regression line shown (a polynomial in this case) provides a good fit of the statistically distributed outliers to the remaining points, with outliers located more than two standard deviations away being excluded from the regression analysis. [Figure 7] FIG. 1 shows a screen capture for GC / MS analysis of volatile organic compounds. [Figure 8] 1 is a flow diagram of one embodiment disclosed herein. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] Components or features that are common to more than one drawing are designated with the same reference number in each of the drawings.

[0014] Referring to Figure 2, there is shown a block diagram of an analytical system 10 that can be used to analyze proteins or other molecules incorporating features of the present invention, as described above. Although the invention will be described with reference to a single embodiment shown in the drawings, it should be understood that the invention can be embodied in many alternative forms or embodiments. In addition, any suitable type of component can be used.

[0015] The analytical system 10 has a sample preparation section 12, another detector section 23, a mass spectrometer section 14, a data analysis system 16, and a computer system 18. The sample preparation section 12 includes a sample introduction unit 20 of a type that introduces a sample containing a protein, peptide, or small molecule drug of interest into the system 10, such as the LCQ Deca XP Max manufactured by Thermo Fisher Scientific, Inc., Waltham, Massachusetts, USA. The sample preparation section 12 may also include an analyte separation unit 22, which is used to perform a preliminary separation of analytes, such as proteins, to be analyzed by the system 10. The analyte separation unit 22 may be one of a chromatography column, an electrophoretic separation unit, such as a gel-based separation unit manufactured by Bio-Rad Laboratories, Inc., Hercules, Calif., or other separation devices, such as ion mobility or pyrolysis, as are known in the art. In electrophoresis, a voltage is applied to the unit to separate proteins as a function of one or more variables, such as migration speed through a capillary, isoelectric focusing point (Hannesh, SM, Electrophoresis 21, 1202-1209 (2000), or by mass (first dimension separation)), or by more than one of these variables, such as by isoelectric focusing and by mass. The latter example is known as two-dimensional electrophoresis.

[0016] The mass spectrometer section 14 can be a conventional mass spectrometer and can be any available, but is preferably one of TOF, quadrupole MS, ion trap MS, qTOF, TOF / TOF, or FTMS. If the mass spectrometer section 14 has an electrospray ionization (ESI) ion source, this type of ion source can also provide the sample input to the mass spectrometer section 14. Generally, the mass spectrometer section 14 can include an ion source 24, a mass analyzer 26 for separating the ions generated by the ion source 24 by mass-to-charge ratio, an ion detector section 28 for detecting the ions from the mass analyzer 26, and a vacuum system 30 for maintaining a vacuum sufficient for the mass spectrometer section 14 to operate most efficiently. If the mass spectrometer section 14 is an ion mobility spectrometer, a vacuum system is generally not required and the data generated is usually called a plasmagram instead of a mass spectrum.

[0017] In parallel with the mass spectrometer section 14 there may be another detector section 23 where a portion of the flow is diverted in a shunt arrangement for near-parallel detection of the sample. This other detector section 23 may be a single channel UV detector, a multi-channel UV spectrometer, or a reflectance (RI) detector, a light scattering detector, a radioactivity monitor (RAM), etc. The RAM may track various metabolites in near real time and correlate them with the mass spectral scan. 14 It is most widely used in drug metabolism studies for C labeling experiments.

[0018] The data analysis system 16 includes a data acquisition section 32 that includes one or a series of analog-to-digital converters (not shown) for converting signals from the ion detector section 28 into digital data. This digital data is provided to a real-time data processing section 34 that processes the digital data by operations such as summing and / or averaging. A post-processing section 36 can be used to perform additional processing of the data from the real-time data processing section 34, including library searching, data storage, and data reporting.

[0019] The computer system 18 provides control of the sample preparation section 12, the mass spectrometer section 14, the other detector section 23, and the data analysis system 16 in a manner described below. The computer system 18 may have a conventional computer monitor or a touch display 40 (or keyboard) that allows for the entry of data and the display of the results of the analyses performed on a suitable screen display. The computer system 18 may be based on any suitable personal computer running, for example, a Windows® or UNIX® operating system, or other suitable operating system. The computer system 18 typically has a hard drive 42 or other type of data storage medium on which the operating system and programs for performing the data analyses described below are stored. A removable data storage device 44 for accepting a CD, floppy disk, memory stick, or other data storage medium is used to load the programs onto the computer system 18. The programs for controlling the sample preparation section 12 and the mass spectrometer section 14 are typically downloaded as firmware for these portions of the system 10. The data analysis system 16 can be a program written in any of a number of programming languages, such as C++, JAVA, or Visual Basic, to carry out the process steps discussed below.

[0020] It should be noted that for more general separations with spectral detection systems to which the present disclosure is applicable, the ion source portion 24 can be replaced with a power source including a light source for a light detection system and an X-ray energy source for an X-ray system. The MS analyzer portion 26 can be replaced with a dispersing device such as a diffraction grating for an optical system with or without a fluorescence option, and the ion detector portion 28 can be replaced with an appropriate corresponding light detector or energy detector.

[0021] In a preferred embodiment, a sample is acquired by the chromatography / mass spectrometry system described in FIG. 2 along with mass spectral data acquired continuously throughout the duration of the run, resulting in a data run as shown in FIG. 3, which is an example of a GC / MS run containing many chromatographic peaks. The detected peaks identified with high confidence through mass spectral library search in the entire sample run are used as naturally occurring "standards" to calibrate the RI for this entire run or other runs under essentially the same chromatographic or other separation conditions, in this case including stationary phase and carrier gas. At least two criteria are used to identify compounds within the run to be used as RI calibration standards. In the first criterion, the quality of the GC / MS search score ("match") is used to tentatively identify each peak. Match values ​​are routinely calculated by comparing the electron impact (EI) ionization fragmentation patterns from the measured mass spectrum with those of pure standards available in EI libraries such as those provided by NIST. The second criterion is the relative retention time of other tentatively identified peaks located before and after the target constituent peak. This can be done in a number of ways, for example, by plotting the known RI values ​​of these compounds of the tentatively identified peaks (available from the NIST database as previously obtained values) as a function of RT and running a least squares or regression analysis through all the tentatively identified compounds. Using this process, a well-spaced subset of the reliably identified compounds can be selected as internal RI calibration standards, thereby accurately calibrating all of the compounds in the entire run for RI. Once calibrated, the RI values ​​of all peaks in the run can be assigned to further improve compound identification, far beyond spectral library searches alone (NIST). The process flow is as follows: a. Automatically detect or otherwise select all peaks in a chromatogram run and optionally apply mixture deconvolution to identify co-eluting compounds as needed, as disclosed in U.S. Provisional Patent Application No. 62 / 632,414, filed February 19, 2018, and International Patent Application No. PCT / US2019 / 018568, published as WO 2019161382. b. Using the summed, averaged, or deconvoluted mass spectra within the RT window of the detected peaks, perform a (e.g., NIST / EPA / NIH or simply called NIST) spectral library search to generate a "Hit List" ranked by search / match quality score. The NIST spectral library is the most widely used public library available, but there are other similar or specialized libraries that can be used, for example, for forensics, or even flavors and fragrances. Home-made and proprietary spectral libraries can also be used. Other spectral libraries such as the Sadtler Database of Infrared Spectra (http: / / www.ir-spectra.com / sadtler / sadtler.htm) can also be used. c. Evaluate each match in the Hit List that is above a specified "good" threshold (e.g., greater than 900 score from the NIST forward search) and optionally evaluate when the difference between the forward search score and the reverse search score is within a specified limit (e.g., 30 or 50 from the NIST search) to create a subset of likely compounds. In the case of co-eluting compounds where the detected chromatographic peak contains more than a single compound mixture causing the forward search score to be impaired due to the presence of spectral interferences, it has been found that the NIST reverse search may recover a very high score by masking or removing the inappropriate m / z values ​​coming from the spectral interferences. Thus, the search score difference between the forward search and the reverse search can be used here as a good indicator of peak purity in selecting likely compounds. d. Plot and perform a least squares regression between the above subset versus the previously obtained RI of each identified probable compound from its RT. Note that multiple mathematical / statistical / numerical models can be used as the regression model, including polynomials, splines, stochastic functions like logit functions, wavelets like Poisson wavelets, or any suitable combination of these either taken over the entire RT range or for a given RT range. Instead of the specific mathematical or statistical models mentioned above, specific graphical models or purely numerical models (e.g., nearest neighbor clustering, permutation, or averaging, etc.) can be used instead. An exhaustive search can be iterated or performed among these possible models to come up with the best possible model, which fits well to the existing data set in a least squares or other sense without overfitting them, i.e., with some residuals of the order of 5-20iu for known retention index measurement error limits. For example, when the polynomial case can be made linear, multiple linear regression (MLR) is readily applied using the methods mentioned in US Pat. Nos. 7,577,538 and 6,983,213. e. Identify "outliers", i.e., previously obtained RI values ​​for compounds that are significantly different from those calculated from the regression model, e.g., based on 1-2 times the standard deviation cutoff or other statistical measure. f. Remove outliers from the above subset used in the regression. g. Repeat the regression analysis on the remaining compounds in the subset until all compounds remaining in the subset are statistically consistent with a pre-set number of statistically acceptable outliers remaining. h. For every compound in the entire run, calculate the RI value of that compound from its RT position using the regression model above. i. Using the RI value match score as an additional metric or filter to further evaluate the likelihood of a possible hit, the calculated RI value is compared to previously obtained RI values ​​from the spectra of the tentatively identified compounds based on the spectral library search or from other libraries, or the RI match value is combined with other metrics, including the spectral match value, into a composite overall search score to more reliably rank the list of tentatively identified compounds for more confident compound identification. The combination of various match scores can take the form of a numerical average after appropriate normalization or weighting, or the search score alone can take the form of a sum of step (or sigmoid) functions that should be (significantly) factored out if the search score is above a certain pre-set threshold.

[0022] Some examples of the process are shown in the figures below. Figure 3 shows a total ion chromatogram (TIC) from a GC / MS analysis of a thermally degraded food flavoring. The sample is complex and contains over 300 compounds, as measured by the mass spectrometer. The temperature programming profile is shown in Figure 4 and includes multiple temperature ramps as well as isothermal sections at the beginning and end of the GC run.

[0023] Each selected or detected peak in the run is searched against the NIST library in an attempt to identify the corresponding compound. In the algorithm used by NIST, the match score indicates the likelihood of correct identification. For example, a match value above 900 is considered a good indicator of correct identification. However, the top match may not necessarily be the correct match due to slight changes in the spectral pattern or possible structural / skeletal / positional isomers resulting from experimental or instrumental variability. However, the top match is correct with a higher degree of certainty.

[0024] The RI data for each compound top identified by the match value is then read from the database and plotted as a function of GC retention time as shown in Figure 5. Visually, trends can be easily recognized in plots with a large number of outliers in the data. Outliers are either 1) compounds that are not correctly identified by the MS search because the compound is not in the library, 2) mixtures of co-eluting compounds whose peaks are misidentified, or 3) the compound's spectrum has low signal-to-noise, which makes the spectral matching worse. These compound outliers are easily removed from the plot (discrimination of purity) by assessing the quality of the match value using the spectral search score and / or the difference between the forward and reverse search scores.

[0025] Once these outliers are identified and removed, the remaining compound RI values ​​can be plotted and fitted to a mathematical / statistical function. As previously mentioned, temperature programming of a GC with multiple temperature ramps causes retention times to deviate non-linearly. To accommodate this deviation from the traditional logarithmic or linear Kovats models, the data are fitted to more flexible functions, such as higher order polynomials as shown in Figure 6.

[0026] However, there is still some possibility that the match value is not always correct. This is especially true for spectrally similar compounds that can have the same chemical formula but different chemical structures (known in the art as various forms of isomers). It is well known that these compounds can be spectrally similar but have significantly different RI values. These outliers can be easily identified as having a statistically significant deviation from the fitted curve. The outliers can be easily identified from the plot shown in Figure 5 and can be removed by the large deviation of the outlier from the fitted curve.

[0027] With outliers detected, the process of fitting the data can be repeated to further remove outliers as necessary. The number of replicates can be set by a reasonable statistical cutoff, for example, allowing 5% of statistical outliers at two times the standard deviation (95% confidence interval under normal distribution). The final fitted curve is the RI calibration that can now be used to assign RI values ​​for each peak and compound in the run given its measured retention time.

[0028] Experimental comparisons of RI values ​​generated by this method are found to be as accurate as those using the traditional method of generating RI calibrations via the more laborious and time-consuming n-alkane external calibration.

[0029] FIG. 7 shows a screen capture for GC / MS analysis of volatile organic compounds. Metrics reported include forward (NIST) and reverse (RevS) library match values, the difference between expected and measured RI values ​​(RIFit, where 100 is a perfect match and values ​​less than that indicate relative deviation from the expected value), and the spectral accuracy match (SA-M) of the compound's molecular ion, as disclosed in U.S. Patent Nos. 6,983,213, 7,493,225, and 7,577,538, as well as U.S. Provisional Patent Application No. 62 / 632,414, filed February 19, 2018, and International Patent Application PCT / US2019 / 018568, published as International Publication WO 2019161382. Compounds that meet the threshold requirements for all metrics are left without a color code, and compounds with one or more of the metrics below the threshold are highlighted in red. This makes it quick and easy for the analyst to review the data and focus only on the "poor" compound identifications. Other intermediate coding colors can also be used to indicate "marginal" identifications.

[0030] FIG. 8 shows the above steps of the flow diagram of the first embodiment described herein, where at 51, during the analyte separation process, spectral data is acquired so that the separated peaks can be detected. At 52, a time window corresponding to the detected peaks with spectral data composited or averaged within the retention time window is selected for searching in the spectral library for a list of possible compounds (hit list). At 53, high quality or high confidence hits from the entire run, including all detected peaks, are selected to form a subset of compounds with previously obtained RI values ​​that are either stored only in the spectral library with spectral data or in a separate database apart from the spectral data. At 54, a regression analysis is performed between the known RI values ​​and the respective RT positions of the known RI values ​​measured for the compounds in the subset. At 55, the RI error between the fitted RI value from the regression model and the known RI value previously available from the spectral search is calculated. At 56, compounds with RI error greater than a preset tolerance or threshold are indicated. At 57, those compounds with large RI errors are then removed from the subset of compounds before the regression is repeated until a majority (e.g., 95%) of the compounds in the subset have their RI errors below the tolerance. At 58, the calculated RI error for any peak in the entire run is used as an additional metric or as part of the overall metric including the spectral search score to rank possible hits for each peak detected or to color code possible compounds indicating the likelihood of correct identification.

[0031] Although the above description contains many details, these should not be construed as limiting the scope of the invention, but merely as providing examples of some possible embodiments. For example, depending on the high reproducibility of a given GC / MS system, a regression model constructed from one sample run can be applied to another sample run under substantially the same separation conditions in its entirety, or only in part by adding additionally identified likely compounds from a feature run into a subset created from a previous run, in order to dynamically enhance and improve the regression model across all runs or over time. Furthermore, the calculated RI values ​​from a sample run can be of sufficient accuracy to be added into a spectral library or database where such values ​​have no or low accuracy. Furthermore, sometimes the compound of interest from a sample run is an unknown true that is not yet included in a spectral library or database, into which the measured spectral data and calculated RI values ​​are added to enhance, augment, or create a new spectral library or database. For mass accuracy and spectral accuracy, as disclosed in U.S. Patent Nos. 7,577,538 and 6,983,213, and for creating, augmenting, or utilizing precise profile mode spectral data and libraries, there are some advantages when acquiring spectral data in raw profile mode and calibrating the profile mode spectral data, as disclosed in U.S. Provisional Patent Application No. 62 / 830,832, filed April 8, 2019, and U.S. Patent Application No. 16 / 843,505, published as US2020-0232956A1. Finally, as disclosed in U.S. Patent Nos. 7,577,538 and 6,983,213, it is possible to determine the elemental composition of unknown compounds not yet included in a library, even using conventional quadrupole mass spectrometers.For unknown compounds, whereby the elemental composition and RI value of the compound are determined, one can search for possible chemical structures using either knowledge of the chemical properties or databases such as ChemSpider (www.chemspider.com). The artificial intelligence (AI) models mentioned above (Matyushin, DD et al., Int. J. Mol. Sci. 22(17), 9194 (2021) or Stein, SE et al., J. Chem. Inf. Model., 47(3), 975-980 (2007)) can be used to predict the RI value of each possible chemical structure, and then the predicted RI can be compared to the calculated RI from the sample run to determine the likelihood that a given structure is or is not a correct hit, potentially providing a turnkey answer machine to the ultimate chemical property question of what the compound is through a single GC / MS experiment.

[0032] Therefore, the scope of the present disclosure should be determined by the appended claims and their legal equivalents, rather than by the examples given. Although the present disclosure has been described with reference to the above-mentioned embodiments, it should be understood that the present disclosure can be embodied in many alternative forms of embodiments. In addition, any suitable size, shape or type of elements or materials can be used. Thus, the present disclosure is intended to embrace all such alternatives, modifications and variations that fall within the scope of the appended claims.

[0033] It will be appreciated that the present disclosure may be embodied in a computer-readable non-transitory storage medium storing computer program instructions that, when executed by a computer system, result in the performance of the steps of the methods described herein. Such storage medium may include any of those mentioned in the description above.

[0034] The techniques described herein are illustrative and should not be construed as implying any particular limitations to the present disclosure. It should be understood that various alternatives, combinations, and modifications may be devised by those skilled in the art. For example, steps associated with a process may be performed in any order unless otherwise specified or required by the step itself. The present disclosure is intended to encompass all such alternatives, modifications, and variations that fall within the scope of the appended claims.

[0035] The term "comprises" or "comprising" should be interpreted as specifying the presence of stated structures, integers, steps or components, but not as excluding the presence of one or more other structures, integers, steps or components, or groups thereof.

Claims

1. 1. A method for analyzing compounds of interest by time-dependent separation when using a spectral detection system, comprising: a. acquiring spectral data for a sample; b. selecting a relevant retention time window for the presence of a possible compound of interest; c) performing a spectral library search using the acquired spectral data within the relevant retention time window to identify possible compounds along with their retention index values; d. using the regression model to perform a regression analysis between the retention index values ​​of the possible compounds and the retention times of the possible compounds by using the subset of identified compounds; e. calculating retention index values ​​based on the measured retention times of the compounds using the regression model; The method comprising:

2. 2. The method of claim 1, wherein the technique for separation is one of gas chromatography (GC), liquid chromatography (LC), supercritical fluid chromatography, ion chromatography (IC), capillary electrophoresis (CE), gel electrophoresis, ion mobility, and pyrolysis.

3. 10. The method of claim 1, wherein the spectral detection system is one of a sector mass spectrometer, a quadrupole mass spectrometer, a time-of-flight (TOF) mass spectrometer, an orbitrap mass spectrometer, a Fourier transform ion cyclotron resonance (FT ICR) mass spectrometer, an optical spectrometer including ultraviolet (UV), visible (Vis), near infrared (NIR), infrared, Raman, fluorescence, X-ray, and X-ray fluorescence.

4. The method of claim 1 , wherein the retention time comprises one of a chromatographic retention time, an elution time, a drift time, and a separation time.

5. 2. The method of claim 1, wherein the retention index values ​​have been previously obtained from measured retention times by using n-alkane related calibration standards in gas chromatography.

6. 2. The method of claim 1, wherein the regression model is one of a polynomial of a given degree, a spline function of a given degree, a stochastic function of a given form, a wavelet of a particular form, a graphical model, a numerical model, and combinations thereof, segmented over a given retention time window.

7. The method of claim 1 , wherein the subset is selected based on a spectral library search quality above a given quality threshold.

8. The method of claim 1 , wherein the subset is selected based on one of a difference between a forward spectrum search and a reverse spectrum search and a combination of a forward spectrum search and a reverse spectrum search.

9. 2. The method of claim 1, wherein the regression model is refined by iteratively removing from the subset those compounds whose calculated retention index values ​​are significantly different from previously obtained retention index values.

10. The method of claim 1 , wherein the regression model is refined by evaluating and selecting from at least one of a mathematical model, a statistical model, a graphical model, and a numerical model.

11. 2. The method of claim 1, wherein the regression model is augmented by further including identified compounds in the existing subset with other sample runs under substantially the same separation conditions.

12. 2. The method of claim 1, wherein the calculated retention index values ​​are compared to retention index values ​​previously obtained for possible compounds identified by said spectral library search.

13. 2. The method of claim 1, wherein the difference between the calculated retention index value and a previously obtained retention index value is used as one of an additional metric and in combination with other metrics to one of filtering or ranking the possible compounds found.

14. The method of claim 13 , wherein one of the other metrics includes a spectral library search score.

15. 2. The method of claim 1, wherein one of the difference between the calculated retention index value and the previously obtained retention index value and a composite search metric that factors said difference is used to color code compounds in a compound hit list on a display where a particular color indicates a particular likelihood of correct identification.

16. 2. The method of claim 1, wherein the regression model is applied to one of the same acquisition and a separate subsequent acquisition under substantially the same separation conditions to calculate a retention index of the compound from the retention time of the compound.

17. 2. The method of claim 1, wherein the regression analysis is multiple linear regression using one of matrix calculation, matrix inversion, singular value decomposition, principal component analysis, and partial least squares.

18. The method of claim 1 , wherein the calculated retention index values ​​for new or additional compounds are added to an existing library or database.

19. The method of claim 1 , wherein the spectral data is acquired in a profile mode, and the subset of compounds is used to one of create and add to a profile mode spectral library.

20. 10. The method of claim 1, wherein the acquired spectral data is calibrated for one of mass accuracy and spectral accuracy to determine elemental composition of unknown compounds not already included in the spectral library.

21. 2. The method of claim 1, wherein the calculated retention indices are compared to retention indices predicted by a computer from possible chemical structures of a given elemental composition to aid in one of determining, selecting, ranking, and screening possible structures of a compound.

22. A spectral detection system including a mass spectrometer operating according to any of the methods of at least one of claims 1 to 21.

23. A computer readable medium having computer readable program instructions readable by the computer for use with a computer associated with a spectral detection system including a mass spectrometer for causing the spectral detection system to operate according to the method of at least one of claims 1 to 21.

Images