ION DETECTION AND PARAMETER ESTIMATION FOR N-DIMENSIONAL DATA [0001] This application claims benefit of and priority to U.S. Provisional Application Serial No. 60/808,901, filed May 26, 2006, the entire contents of which is hereby incorporated by reference in its entirety. FIELD OF THE INVENTION [0002] The present invention relates generally to analysis of compounds, and, more particularly, to detection and quantification of ions collected by liquid chromatography, ion-mobility spectrometry and mass spectrometry. BACKGROUND OF THE INVENTION [0003] Mass spectrometers (MS) are used widely for identifying and quantifying molecular species in a sample. During analysis, molecules from the sample are ionized to form ions that are introduced into the mass spectrometer for analysis. The mass spectrometer measures the mass-to-charge ratio (m/z) and intensity of the introduced ions. [0004] Mass spectrometers are limited in the number of different ions reliably detected and quantified within a single sample spectrum. As a result, samples containing many molecular species may produce spectra that are too complex for interpretation or analysis using conventional mass spectrometers. [0005] In addition, the concentration of molecular species often varies over a wide range. Biological samples, for example, typically have a greater number of molecular species at lower concentrations than at higher concentrations. Thus, a significant fraction of ions appear at low concentration. The low concentration is often near the detection limit of common mass spectrometers. Moreover, at low concentration, ion detection suffers from background noise and/or interfering background molecules. Consequently, detection of such low abundance species can be improved by removing as much of the background noise as possible and reducing the number of interfering species that are present in the spectrum. [0006] A chromatographic separation, prior to injecting the sample into the mass spectrometer, is commonly used to reduce the complexity of such spectra. For example, peptides or proteins often produce clusters of ions that elute at a common chromatographic retention time and thus produce peaks that overlap in a spectrum. Separating the clusters from the different molecules, in time, helps to simplify interpretation of the spectra produced by such clusters. [0007] Common chromatographic separation instruments include gas chromatographs (GC) and liquid chromatographs (LC). When coupled to a mass spectrometer, the resulting systems are referred to as GC/MS or LC/MS systems. GC/MS or LC/MS systems are typically on-line systems in which the output of the GC or LC is coupled directly to the MS. [0008] A combined LC/MS system provides an analyst with a powerful means to identify and to quantify molecular species in a wide variety of samples. Common samples contain a mixture of a few or thousands of molecular species. The molecules often exhibit a wide range of properties and characteristics, and each molecular species can yield more than one ion. For example, the mass of a peptide depends on the isotopic forms of its nucleus, and an electrospray interface can ionize peptides and proteins into families of charge states. [0009] In an LC/MS system, a sample is injected into the liquid chromatograph at a particular time. The liquid chromatograph causes the sample to elute over time resulting in an eluent that exits the liquid chromatograph. The eluent exiting the liquid chromatograph is continuously introduced into the ionization source of the mass spectrometer. As the separation progresses, the composition of the mass spectrum generated by the MS evolves and reflects the changing composition of the eluent. [0010] Typically, at regularly spaced time intervals, a computer-based system samples and records the spectrum. In conventional systems, the acquired spectra are analyzed after completion of the LC separation. [0011] After acquisition, conventional LC/MS systems generate one-dimensional spectra and chromatograms. The response (or intensity) of an ion is the height or area of the peak as seen in either the spectrum or the chromatogram. To analyze spectra or chromatograms generated by conventional LC/MS systems, peaks in such spectra or chromatograms that correspond to ions must be located or detected. The detected peaks are analyzed to determine properties of the ions giving rise to the peaks. These properties include retention time, mass-to-charge ratio and intensity. [0012] Mass or mass-to-charge ratio (m/z) estimates for an ion are derived through examination of a spectrum that contains the ion. Retention time estimates for an ion are derived by examination of a chromatogram that contains the ion. The time location of a peak apex in a single mass-channel chromatogram provides an ion's retention time. The m/z location of a peak apex in a single spectral scan provides the ion's m/z value. [0013] A conventional technique for detecting ions using an LC/MS system forms a total ion chromatogram (TIC). Typically, this technique is applied if there are relatively few ions requiring detection. A TIC is generated by summing, within each spectral scan, all responses collected over all m/z values and plotting the sums against scan time. Ideally, each peak in a TIC corresponds to a single ion. [0014] Co-elution of peaks from multiple molecules is one possible problem with this method of detecting peaks in a TIC. As a result of co-elution, each isolated peak seen in the TIC may not correspond to a unique ion. A conventional method for isolating such co-eluted peaks is to select the apex of one peak from the TIC and collect spectra for the time corresponding to the selected peak's apex. The resulting spectral plot is a series of mass peaks, each presumably corresponding to a single ion eluting at a common retention time. [0015] For complex mixtures, co-elution also typically limits summing of spectral responses to sums only over a subset of collected channels, e.g., by summing over a restricted range of m/z channels. The summed chromatogram provides information about ions detected within the restricted m/z range. In addition, spectra can be obtained for each chromatographic peak apex. To identify all ions in this manner, multiple summed chromatograms are generally required. [0016] Another difficulty encountered with peak detection is detector noise. A common technique for mitigating detector noise effects is to signal-average spectra or chromatograms. For example, the spectra corresponding to a particular chromatographic peak can be co-added to reduce noise effects. Mass-to-charge ratio values as well as peak areas and heights can be obtained from analyzing the peaks in the averaged spectrum. Similarly, co-adding chromatograms centered on the apex of a spectral peak can mitigate noise effects in chromatograms and provide more accurate estimates of retention time as well as chromatographic peak areas and heights. [0017] Aside from these problems, additional difficulties are encountered when conventional peak detection routines are used to detect chromatographic or spectral peaks. If performed manually, such conventional methods are both subjective and tedious. When performed automatically, such methods can still be subjective, due to a subjective selection of thresholds for identification of peaks. Further, these conventional methods tend to be inaccurate because they analyze data using only a single extracted spectrum or chromatogram, and do not provide ion parameter estimates having the highest statistical precision or lowest statistical variance. Finally, conventional peak-detection techniques do not necessarily provide uniform, reproducible results for ions at low concentration; or for complex chromatograms, where co-elution and ion interference tend to be common problems. BRIEF SUMMARY OF THE INVENTION [0018] Some embodiments of the present invention involve analysis instrumentation and methods that entail data of three or more dimensions. For example, some preferred embodiments of the invention involve apparatus that include LC, ion-mobility spectrometry (IMS) and MS. Some aspects of the invention arise from the realization that LC/IMS/MS and other higher dimensional data-generating techniques benefit from efficient data evaluation through use of convolution filters to reduce artifacts caused by noise and/or peak interference. Moreover, temporarily collapsing data in a relatively low, or lowest dimension, such as the ion-mobility dimension, speeds analysis, and allows a higher or highest resolution dimension, such as the mass-spectrometry dimension to identify large portions of collected data that can be ignored during data analysis. Further, the ion-mobility dimension, for example, supports distinguishing of ion peaks that otherwise overlap in other dimensions, such as a retention time and/or mass dimension. [0019] For example, in some LC/IMS/MS-base embodiments, faster, more efficient characterization of data provides ion parameters, such as ion mobility, mass-to-charge ratio (m/z), retention time and ion intensity, which are accurately and optimally estimated by creating a data matrix of data collapsed in the ion-mobility dimension, and convolved with a fast, linear two-dimensional finite impulse response (FIR) filter to generate an output convolved matrix. A peak detection routine is applied to the output convolved matrix to identify peaks corresponding to ions in the sample. [0020] The identified peaks are optionally used to indicate which portions of the original data contain peaks. These indicated portions are optionally convolved with a three-dimensional filter, and ions peaks are then located in all dimensions. Evaluation of the low-resolution dimension supports, for example, elucidation of ion peaks that are overlapped in other dimensions. [0021] Accordingly, in one illustrative embodiment, the invention features a method of LC/IMS/MS analysis. The method includes obtaining noisy raw data from a sample. The data includes a set of three-dimensional data elements each associating an ion-count intensity with a retention-time dimension, an ion-mobility dimension, and a mass-to-charge-ratio dimension, where the noise is associated with ion-peak artifacts. The method further includes: collapsing, in the ion mobility dimension, the set of data elements to obtain a set of collapsed data elements each associating a combined ion-count intensity with the retention-time and mass-to-charge-ratio dimensions; convolving the set of collapsed data elements with an artifact-reducing filter, the filter associated with a two-dimensional matrix, thereby obtaining a convolved set of data elements having reduced peak artifacts; locating, in the retention-time and mass-to-charge-ratio dimensions, ion peaks of the convolved collapsed set of data elements; selecting one or more portions of the raw data for further analysis, in response to the locations of the ion peaks of the convolved set of data elements; and locating, at least in the ion mobility dimension, one or more ion peaks for each of the portions of raw data. [0022] In a second illustrative embodiment, the invention features a method of N-dimensional analysis. The method includes: obtaining noisy data from a sample, the data including a set of data elements each associating an ion-count intensity with at least three dimensions of differing resolution; convolving the set of data elements with an artifact-reducing filter to produce a convolved set of data elements; and locating one or more ion peaks in the convolved set of data elements. [0023] In another aspect, the invention involves chemical processing apparatus. The apparatus includes a control unit that is configured to implement, for example, one of the above-described methods. BRIEF DESCRIPTION OF THE DRAWINGS [0024] Figure 1 is a schematic diagram of an exemplary LC/MS system according to an embodiment of the present invention. [0025] Figure 2 is a diagram of an exemplary chromatographic or spectral peak. [0026] Figure 3 illustrates exemplary spectra for three ions produced during an exemplary LC/MS experiment for three times. [0027] Figure 4 illustrates chromatograms corresponding to the exemplary ions of Figure 3. [0028] Figure 5 is a flow chart for a method for processing data according to an embodiment of the present invention. [0029] Figure 6 is a graphical flow chart for a method for processing data according to an embodiment of the present invention. [0030] Figure 7 is a graphical flow chart for a method for determining thresholds for use in detecting ions according to an embodiment of the present invention. [0031] Figure 8 illustrates an exemplary data matrix according to an embodiment of the present invention. [0032] Figure 9 illustrates a contour plot representation of an exemplary data matrix created from the data of Figures 3 and 4 according to an embodiment of the present invention. [0033] Figure 10 is a flow chart for a simplified method of processing data in the absence of noise according to an embodiment of the present invention. [0034] Figure 11 illustrates an effect of a co-eluting ion on the exemplary data matrix of Figure 9. [0035] Figure 12 illustrates a "shoulder" effect of a co-eluting ion on the exemplary data illustrated in Figure 3. [0036] Figure 13 illustrates how noise affects exemplary data in a data matrix created according to embodiments of the present invention. [0037] Figure 14A illustrates spectra for three ions corresponding to the exemplary data illustrated in the data matrix shown in Figure 13. [0038] Figure 14B illustrates chromatograms for three ions corresponding to the exemplary data illustrated in the data matrix shown in Figure 13. [0039] Figure 15 illustrates an exemplary one-dimensional apodized Savitzky-Golay second-derivative filter according to an embodiment of the present invention. [0040] Figure 16A illustrates the cross section of an exemplary one-dimensional filter in the spectral (m/z) direction according to an embodiment of the present invention. [0041] Figure 16B illustrates the cross section of an exemplary one-dimensional filter in the chromatographic (time) direction according to an embodiment of the present invention. [0042] Figure 16C illustrates the cross section of an exemplary one-dimensional smoothing filter f1 in the spectral (m/z) direction according to an embodiment of the present invention. [0043] Figure 16D illustrates the cross section of an exemplary one-dimensional second derivative filter g1 in the chromatographic direction according to an embodiment of the present invention. [0044] Figure 16E illustrates the cross section of an exemplary one-dimensional smoothing filter g2 in the chromatographic direction according to an embodiment of the present invention. [0045] Figure 16F illustrates the cross section of an exemplary one-dimensional second-derivative filter f2 in the spectral (m/z) direction according to an embodiment of the present invention. [0046] Figure 17A illustrates an exemplary peak that can be generated by LC/MS data as stored in a data matrix according to embodiments of the present invention. [0047] Figure 17B illustrates a point-source response (finite impulse response) of an exemplary rank-2 filter according to an embodiment of the present invention. [0048] Figure 17C illustrates a simulation of two LC/MS peaks having equal mass and that are nearly, but not identically, co-incident in time. [0049] Figure 17D illustrates the peak cross section in mass of the two-peak simulation of Figure 17C. [0050] Figure 17E illustrates the peak cross section in time of the two-peak simulation of Figure 17C. [0051] Figure 17F illustrates the effect of adding counting (shot) noise to the two-peak simulation of Figure 17C. [0052] Figure 17G illustrates the peak cross section in mass of the added noise two-peak simulation of Figure 17F. [0053] Figure 17H illustrates the peak cross section in time of the added noise two-peak simulation of Figure 17F. [0054] Figure 171 illustrates the result convolving a rank-2 filter with simulated data of Figure 17F. [0055] Figure 17J illustrates the peak cross section in mass of the result illustrated in Figure 171. [0056] Figure 17K illustrates the peak cross section in time of the result illustrated in Figure 171. [0057] Figure 18 illustrates a flow chart for performing real-time processing of data according to an embodiment of the present invention. [0058] Figure 19 is a graphical illustration of a method for performing realtime processing of a data according to the method of the flow chart of Figure 18. [0059] Figure 20 is a flow chart for a method for determining appropriate thresholds according to an embodiment of the present invention. [0060] Figure 21 is a flow chart for a method for determining a peak purity metric according to an embodiment of the present invention. [0061] Figure 22A illustrates an exemplary LC/MS data matrix resulting from two parent molecules and the resulting multiplicity of molecules. [0062] Figure 22B illustrates an exemplary complex spectrum corresponding to the data of Figure 22A at a time t1. [0063] Figure 22C illustrates an exemplary complex spectrum corresponding to the data of Figure 22A at a time t2. [0064] Figure 23 is a graphical chart illustrating how related ions can be identified in the unmodified and modified ion lists generated by an embodiment of the present invention. [0065] Figure 24 is a flow diagram of a method of analysis, in accordance with one embodiment of the invention. DETAILED DESCRIPTION OF THE INVENTION DEFINITIONS [0066] "Chromatography" - refers to equipment and/or methods used in the separation of chemical compounds. Chromatographic equipment typically moves fluids and/or ions under pressure and/or electrical and/or magnetic forces. The word "chromatogram," depending on context, herein refers to data or a representation of data derived by chromatographic means. A chromatogram can include a set of data points, each of which is composed of two or more values; one of these values is often a chromatographic retention time value, and the remaining value(s) are typically associated with values of intensity or magnitude, which in turn correspond to quantities or concentrations of components of a sample. [0067] The invention supports the generation and analysis of chromatographic data. Some embodiments of the invention involve instruments that include a single module that separates sample compounds while other embodiments involve multiple modules. For example, principles of the invention are applicable to liquid chromatography apparatus as well as to, for example, apparatus that include liquid chromatography, ion-mobility spectrometry and mass spectrometry modules. In some multi-module-based embodiments, a chromatographic module is placed in fluidic communication with an ion-mobility spectrometry module through an appropriate interface; the IMS module is, in turn, interfaced to a mass-spectrometry module through use of an appropriate interface, such as an electrospray-ionization interface. Some appropriate interfaces at times create or maintain separated materials in an ionic form. A stream of sample fluid is typically vaporized, ionized, and delivered to an inlet orifice of a mass-spectrometry module. [0068] Thus, some embodiments produce multi-dimensional data composed of sets of data elements, each of which has valves associated with measurement dimensions such as retention time (derived from a chromatography module), ion mobility and mass-to-charge ratio. A unique set of dimensional valves is experimentally linked to, for example, a valve of ion intensity as measured in a mass spectrometry module. [0069] Protein - herein refers to a specific primary sequence of amino acids assembled as a single polypeptide. [0070] Peptide - herein refers to a specific sequence of amino acids assembled as a single polypeptide contained within the primary sequence of a protein. [0071] Precursor peptides - tryptic peptides (or other protein cleavage products) that are generated using a protein-cleavage protocol. The precursors are optionally separated chromatographically and passed to a mass spectrometer. An ion source ionizes these precursor peptides to typically produce a positively charged, protenated form of the precursor. The mass of such positively charged protenated precursor ion is herein referred as the "mwHPIus" or "MH+" of the precursor. In the following, the term "precursor mass" refers generally to the protenated, mwHPIus or MH+ mass of the ionized, peptide precursor. [0072] Fragments — Multiple types of fragments can occur in LC/MS analyses. In the case of tryptic peptide precursors, fragments can include polypetide ions that are produced from collisional fragmentation of the intact peptide precursors and whose primary amino acid sequence is contained within the originating precursor peptide. Y-ions and B-ions are examples of such peptide fragments. Fragments of tryptic peptides can also include immonium ions, functional groups such as a phosphate ion (P03), mass tags cleaved from a specific molecule or class of molecules, or "neutral loss" of water (H20) or ammonia (NH3) molecules from the precursor. [0073] Y-ions and B-ions - If a peptide fragments at the peptide bond, and if a charge is retained on the N terminal fragment, that fragment ion is termed a B-ion. If the charge is retained on the C terminal fragment, the fragmention is termed a Y-ion. A more comprehensive list of possible fragments and their nomenclature is provided in Roepstorff and Fohlman, Biomed Mass Spectrom, 1984; 11(11):601 and Johnson et al, Anal. Chem 1987, 59(21): 2621:2625. [0074] Retention time - in context, typically refers to the point in a chromatographic profile at which an entity reaches its maximum intensity. [0075] Ions - a peptide, for example, typically appears in an LC/MS analysis as an ensemble of ions due to the natural abundance of the isotopes of the constituent elements. An ion has, for example, a retention time and an m/z value. The mass spectrometer (MS) detects only ions. The LC/MS technique produces a variety of observed measurements for every detected ion. This includes: the mass-to-charge ratio (m/z), mass (m), the retention time, and the signal intensity of the ion, such as a number of ions counted. [0076] Noise - used herein to refer to a raw-data component arising from sources such as detector noise, including Poisson noise due to counting statistics and Gaussian, Johnson noise due to thermal effects, and other noise sources that tend to hide real ion peaks or produce false ion peaks. [0077] Artifact - refers herein to false peaks in raw data, as arise from, for example, noise, peak interference and peak overlap. [0078] Generally, an LC/IMS/MS analysis optionally provides an empirical description of, for example, a peptide in terms of its mass, charge, retention time, mobility and total intensity. When a peptide elutes from a chromatographic column, it elutes over a specific retention time period and reaches its maximum signal at a single retention time. After ionization and (possible) fragmentation, the peptide appears as a related set of ions. The different ions in the set correspond to different isotopic compositions and charges of the common peptide. Each ion within the related set of ions produces a single peak retention time and peak shape. Since these ions originate from a common peptide, the peak retention time and peak shape of each ion is identical, within some measurement tolerance. The MS acquisition of each peptide produces multiple ion detections for all isotopes and charge states, all sharing the same peak retention-time and peak shape within some measurement tolerance. [0079] In an LC/MS separation, a single peptide (precursor or fragment) produces many ion detections, which appear as a cluster of ions, having multiple charge states. Deconvolution of these ion detections from such a cluster, indicates the presence of a single entity of a unique monoisotopic mass, at a specific retention time, of a measured signal intensity, in a charge state. [0080] Embodiments of the present invention can be applied to a variety of applications including large-molecule, non-volatile analytes that can be dissolved in a solvent. Although embodiments of the present invention are described hereinafter with respect to LC, LC/MS or LC/IMS/MS systems, embodiments of the present invention can be configured for operation with other analysis techniques, including GC, GC/MS and GC/IMS/MS systems. For context, embodiments that utilize 1-D and 2-D matrices for analysis of LC/MS data are described first, with reference to Figures 1-23. Subsequently, some preferred embodiments of the present invention, relating to LC/IMS/MS and higher dimensional techniques, are described with reference to Figures 24. [0081] Figure 1 is a schematic diagram of an exemplary LC/MS system 101 according to an embodiment of the present invention. LC/MS analysis is performed by automatically or manually injecting a sample 102 into a liquid chromatograph 104. A high pressure stream of chromatographic solvent provided by pump 103 and injector 105 forces sample 102 to migrate through a chromatographic column 106 in liquid chromatograph 104. Column 106 typically comprises a packed column of silica beads whose surface comprises bonded molecules. Competitive interactions between the molecular species in the sample, the solvent and the beads determine the migration velocity of each molecular species. [0082] A molecular species migrates through column 106 and emerges, or elutes, from column 106 at a characteristic time. This characteristic time commonly is referred to as the molecule's retention time. Once the molecule elutes from column 106, it can be conveyed to a detector, such as a mass spectrometer 108. [0083] A retention time is a characteristic time. That is, a molecule that elutes from a column at retention time t in reality elutes over a period of time that is essentially centered at time t. The elution profile over the time period is referred to as a chromatographic peak. The elution profile of a chromatographic peak can be described by a bell-shaped curve. The peak's bell shape has awidth that typically is described by its full width at half height, or half-maximum (FWHM). The molecule's retention time is the time of the apex of the peak's elution profile. Spectral peaks appearing in spectra generated by mass spectrometers have a similar shape and can be characterized in a similar manner. Figure 2 illustrates an exemplary chromatographic or spectral peak 202 having a peak apex 204. The FWHM and height or the peak 202 are also illustrated in Figure 2. [0084] For purposes of subsequent description, peaks are assumed to have a Gaussian profile as shown in Figure 2. For a Gaussian profile, the FWHM is approximately 2.35 times the standard deviation crof the Gaussian profile. [0085] Chromatographic peak width is independent of peak height and is substantially a constant characteristic of a molecule for a given separation method. In the ideal case, for a given chromatographic method all molecular species would elute with the same peak width. However, typically peak widths change as a function of retention time. For example, molecules that elute at the end of a separation can display peak widths that are several times wider than peak widths associated with molecules that elute early in the separation. [0086] In addition to its width, a chromatographic or spectral peak has a height or area. Generally, the height and area of the peak are proportional to the amount or mass of the species injected into the liquid chromatograph. The term intensity commonly refers to either the height or area of the chromatographic or spectral peak. [0087] Although chromatographic separation is a substantially continuous process, detectors analyzing the eluent typically sample the eluent at regularly spaced intervals. The rate at which a detector samples the eluent is referred to as the sample rate or sample frequency. Alternatively, the interval at which a detector samples the eluent is referred to as the sampling interval or sample period. Because the sample period must be long enough so that the system adequately samples the profile of each peak, the minimum sample period is limited by the chromatographic peak width. As an example, the sample period can be set so that approximately five (5) measurements are made during the FWHM of a chromatographic peak. [0088] In an LC/MS system, the chromatographic eluent is introduced into a mass spectrometer (MS) 108 for analysis as shown in Figure 1. MS 1Q8 comprises a desolvation system 110, an ionizer 112, a mass analyzer 114, a detector 116, and a computer 118. When the sample is introduced into MS 108, desolvation system 110 removes the solvent, and ionizing source 112 ionizes the analyte molecules. Ionization methods to ionize molecules that evolve from LC 104 include electron-impact (El), electrospray (ES), and atmospheric chemical ionization (APCI). Note that in APCI, the order of ionization and desolvation is reversed. [0089] The ionized molecules are then conveyed to mass analyzer 114. Mass analyzer 114 sorts or filters the molecules by their mass-to-charge ratio. Mass analyzers, such as mass analyzer 114 that are used to analyze ionized molecules in MS 108 include quadrupole mass analyzers (Q), time-of-flight (TOF) mass analyzers, and Fourier-transform-based mass spectrometers (FTMS). [0090] Mass analyzers can be placed in tandem in a variety of configurations, including, e.g., quadrupole time-of-flight (Q-TOF) mass analyzers. A tandem configuration enables on-line collision modification and analysis of an already mass-analyzed molecule. For example, in triple quadrupole based massed analyzers (such as Q1-Q2-Q3 or Q1-Q2-TOF mass analyzers), the second quadrupole (Q2), imports accelerating voltages to the ions separated by the first quadrupole (Q1). These ions, collide with a gas expressly introduced into Q2. The ions fragment as a result of these collisions. Those fragments are further analyzed by the third quadrupole (Q3) or by the TOF. Embodiments of the present invention are applicable to spectra and chromatograms obtained from any mode of mass-analysis such as those described above. [0091] Molecules at each value for m/z are then detected with detection device 116. Exemplary ion detection devices include current measuring electrometers and single ion counting multi-channel plates (MCPs). The signal from an MCP can be analyzed by a descriminator followed by a time-domain-converter (TDC) or by an analog to digital (ATD) converter. For purposes of the present description, an MCP detection-based system is assumed. As a result, detector response is represented by a specific number of counts. This detector response {i.e., number of counts) is proportional to the intensity of ions detected at each mass-to-charge-ratio interval. [0092] An LC/MS system outputs a series of spectra or scans collected over time. A mass-to-charge spectrum is intensity plotted as a function of m/z. Each element, a single mass-to-charge ratio, of a spectrum is referred to as a channel. Viewing a single channel over time provides a chromatogram for the corresponding mass-to-charge ratio. The generated mass-to-charge spectra or scans can be acquired and recorded by computer 118 and stored in a storage medium such as a hard-disk drive that is accessible to computer 118. Typically, a spectrum or chromatogram is recorded as an array of values and stored by computer system 118. The array can be displayed and mathematically analyzed. [0093] The specific functional elements that make up an MS system, such as MS 108, can vary between LC/MS systems. Embodiments of the present invention can be adapted for use with any of a wide range of components that can make up an MS system. [0094] After chromatographic separation and ion detection and recordation, the data is analyzed using a post-separation data analysis system (DAS). In an alternate embodiment of the present invention, the DAS performs analysis in real-time or near real-time. The DAS is generally implemented by computer software executing on a computer such as computer 118 shown in Figure 1. Computers that can be configured to execute the DAS as described herein are well known to those skilled in the art. The DAS is configured to perform a number of tasks, including providing visual displays of the spectra and/or chromatograms as well as providing tools for performing mathematical analysis on the data. The analyses provided by the DAS include analyzing the results obtained from a single injection and/or the results obtained from a set of injections to be viewed and further analyzed. Examples of analyses applied to a sample set include the production of calibration curves for anaiytes of interest, and the detection of novel compounds present in the unknowns, but not in the controls. A DAS according to embodiments of the present invention is described herein. [0095] Figure 3 illustrates exemplary spectra for three ions (ion 1, ion 2 and ion 3) produced during an exemplary LC/MS experiment. Peaks associated with ion 1, ion 2 and ion 3 appear within a limited range of retention time and m/z. For the present example, it is assumed that the mass-to-charge ratios of ion 1, ion 2 and'ion 3 are different, and that the molecular parents of the ions eluted at nearly, but not exactly, identical retention times. As a result, the elution profiles of the respective molecules overlap or co-elute. Under these assumptions, there is a time when all three molecules are present in the ionizing source of the MS. For example, the exemplary spectrum illustrated in Figure 3 were collected when all three ions were present in the MS ionization source. This is apparent because each spectrum exhibits a peak associated with each of ions 1, 2 and 3. As can be seen in the exemplary spectra illustrated in Figure 3, there is no overlap of spectral peaks. The lack of overlap indicates that the mass spectrometer resolved these spectral peaks. The location of the apex of the peaks corresponding to each of ions 1, 2 and 3 represents its mass-to-charge ratio. [0096] It is not possible to determine precise retention times or even relative retention times at which ions in a spectrum elute using only a single spectrum. For example, it can be seen that at the time the data for Spectrum B was collected, all three molecules associated with ions 1, 2 and 3 were eluting from the column. However, analyzing only Spectrum B, it is not possible to determine a relationship between the elution times of ions 1, 2 and 3. Thus, Spectrum B could have been collected at a time corresponding to the beginning of a chromatographic peak, as the molecule began to elute from the column, or from the end of the chromatographic peak, when the molecule was nearly finished eluting or at some time in between. [0097] More accurate information related to retention time can be obtained by examining successive spectra. This additional information can include the retention time of the eluting molecules or at least the elution order. For example, assume Spectra A, B, and C shown in Figure 3 were collected successively such that Spectrum A was collected at time tA; Spectrum B was collected at a later time tB; and Spectrum C was collected at time tC, which is a time later than time tB. Then, the elution order of the respective molecules can be determined by examining the relative heights of the peaks appearing in spectra successively collected as time progresses from tA to tC. Such examination reveals that as time progresses ion 2 decreases in intensity relative to ion 1, and that ion 3 increases in intensity relative to ion 1. Therefore, ion 2 elutes before ion 1, and ion 3 elutes after ion 1. [0098] This elution order can be verified by generating chromatograms corresponding to each peak found in a spectrum. This can be accomplished by obtaining the m/z value at the apex of each of the peaks corresponding to ions 1, 2 and 3. Given these three m/z values, the DAS extracts from each spectrum the intensity obtained at that m/z for each scan. The extracted intensities are then plotted versus elution time. Such a plot is illustrated in Figure 4. It can be seen that the plots in Figure 4 represent the chromatograms for ions 1, 2, and 3 at the m/z values obtained by examining the peaks in Figure 3. Each chromatogram contains a single peak. Examination of the chromatograms for ions 1, 2 and 3 as illustrated in Figure 4 confirms that ion 2 elutes at the earliest time and that ion 3 elutes at the latest time. The apex location in each of the chromatograms shown in Figure 4 represents the elution time for the molecule corresponding to the respective ions. [0099] With this introduction in mind, embodiments of the present invention relate to analyzing experimental analysis outputs, such as spectra and chromatograms, to optimally detect ions and quantify parameters related to the detected ions. Moreover, embodiments of the present invention can provide significantly simplified spectra and chromatograms. [00100] Figure 5 is a flow chart 500 for processing experimental analysis output, such as spectra and chromatograms. Flow chart 500 can be embodied in a number of ways including in the DAS described above. In the embodiment of the present invention illustrated in Figure 5, analysis proceeds as follows: STEP 502: Create a two-dimensional data matrix having chromatographic and spectral data. STEP 504: Specify a two-dimensional convolution filter to apply to the data matrix. STEP 506: Apply the two-dimensional convolution filter to the data matrix. For example, the data matrix can be convolved with the two-dimensional filter. STEP 508: Detect peaks in the output of the application of the two-dimensional filter to the data matrix. Each detected peak is deemed to correspond to an ion. Thresholding can be used to optimize peak detection. STEP 510: Extract ion parameters for each detected peak. The parameters include ion characteristics such as retention time, mass-to-charge ratio, intensity, peak width in the spectral direction and/or peak width in the chromatographic direction. STEP 512: Store the ion parameters associated with extracted ions in a list or table. Storage can be performed as each peak is detected or after a plurality or all of the peaks have been detected. STEP 514: Use the extracted ion parameters to post-process the data. For example, the ion parameter table can be used to simplify the data. Such simplification can be accomplished, for example, by windowing to reduce spectral or chromatographic complexity. Properties of the molecules can be inferred from the simplified data. [00101] Figures 6 and 7 are graphical flow charts describing the foregoing steps of flow chart 500. Figure 6 is a graphical flowchart 602 of a method for processing LC/MS data according to an embodiment of the present invention. More particularly, each element of graphical flowchart 602 illustrates the result of a step according to an embodiment of the present invention. Element 604 is an exemplary LC/MS data matrix created according to an embodiment of the present invention. As described below, the LC/MS data matrix can be created by placing LC/MS spectra collected at successive times in successive columns of a data matrix. Element 606 is an exemplary two-dimensional convolution filter that can be specified according to desired filtering characteristics. Considerations for specifying the two-dimensional filter are described in more detail below. Element 608 represents application of the two-dimensional filter of element 606 to the LC/MS data matrix of element 604 according to an embodiment of the present invention. An exemplary such application of the two-dimensional filter to the LC/MS data matrix is a two-dimensional convolution wherein the LC/MS data matrix is convolved with the two-dimensional convolution filter. The output of the filtering step is the output data matrix, an example of which is illustrated as element 610. Where the application of the filter to the data matrix comprises a convolution, the output is an output convolved matrix. {00102] Element 612 illustrates an exemplary result of performing peak detection on the output data matrix to identify or detect peaks associated with ions. Thresholding can be used to optimize the peak detection. At this point, the ions are considered detected. Element 614 is an exemplary list or table of the ion properties created using the detected ions. [00103] Figure 7 is a graphical flowchart 702 illustrating results of determining a detection threshold and its application to further consolidate the ion parameter table according to an embodiment of the present invention. Element 706 represents exemplary peak data accessed from the ion parameter list, Element 704. Element 706 illustrate results of determining a detection threshold using the accessed data. The determined threshold is applied to the ion parameter list generated as Step 704 to generate an edited ion parameter list an emample of which is illustrated as Step 708. The foregoing steps are now explained in more detail. Step 1: Create data matrix [00104] Rather than view the output of an LC/MS analysis as distinct series of spectra and chromatograms, it is advantageous to configure the LC/MS output as a data matrix of intensities. In an embodiment of the present invention, the data matrix is constructed by placing the data associated with each successive spectrum collected at increasing time in successive columns of a data matrix thereby creating a two-dimensional data matrix of intensities. Figure 8 illustrates an exemplary such data matrix 800 in which five (5) spectra successively collected in time are stored in successive columns 801-805 of data matrix 800. When the spectra are stored in this manner, the rows of data matrix 800 represent chromatograms at corresponding m/z values in the stored spectra. These chromatograms are indicated by rows 811-815 in data matrix 800. Thus, in matrix form, each column of the data matrix represents a spectrum collected at a particular time, and each row represents a chromatogram collected at a fixed m/z. Each element of the data matrix is an intensity value collected at a particular time (in the corresponding chromatogram) for a particular m/z (in the corresponding spectrum). Although the present disclosure assumes column-oriented spectral data and row-oriented chromatographic data, in alternate embodiments of the present invention, the data matrix is oriented such that rows represent spectra and columns represent chromatograms. [00105] Figure 9 is an exemplary graphical representation (in particular, a contour plot) of a data matrix generated as described above by storing spectral data in successive columns of the data matrix. In the contour plot illustrated in Figure 9, each of ions 1, 2 and 3 appears as an island of intensity. The contour plot distinctly shows not only the presence of the three ions, but also that the elution order is ion 2, followed by ion 1, followed, by ion 3. Figure 9 also shows three apices 902a, 902b and 902c. Apex 902a corresponds to ion 1, apex 902b corresponds to ion 2 and apex 902c corresponds to ion 3. The locations of apices 902a, 902b, and 902c correspond to the m/z and retention time for ions 1, 2 and 3 respectively. The height of the apex above the zero value floor of the contour plot is a measure of the-ion's intensity. The counts or intensities associated with a single ion are contained within an ellipsoidal region or island. The FWHM of this region in the m/z (column) direction is the FWHM of the spectral (mass) peak. The FWHM of this region in the row (time) direction is the FWHM of the chromatographic peak. [00106] The innermost of the concentric contours forming an island identifies the element having the highest intensity. This local maximum or maximal element has an intensity greater than its nearest neighbors. For example, for two-dimensional data contours, a local maximum or apex is any point whose amplitude is greater than its nearest-neighbor elements. In one embodiment of the present invention, a local maximum or apex must be greater than eight (8) nearest neighbor elements. For example in the Table 1, the central element is a local maximum because each of the 8 adjoining elements has a value less than 10. (TABLE REMOVED) Table 1: Example showing maximum [00107] There are six lines drawn through the contour plot of Figure 9. The three horizontal lines, labeled ion 1, ion 2 and ion 3, identify cross sections corresponding to the chromatograms for ions 1, 2 and 3 respectively as illustrated in Figure 4. The three vertical lines, labeled A, B and C, identify cross sections corresponding to the mass spectra 3A, 3B and 3C respectively as illustrated in Figure 3. [00108] After the data matrix is created, ions are detected. For each detected ion, ion parameters, such as retention time, m/z and intensity, are obtained. If the data matrix is free of noise and if the ions do not interfere with one another (e.g., by chromatographic co-elution and spectral interference), then each ion produces a unique, isolated island of intensity, as illustrated in the contour plot of Figure 9. [00109] As shown in Figure 9, each island contains a single maximal element. Where there is no noise, co-elution or interference, ion detection and parameter quantification according to an embodiment of the present invention proceeds as follows as shown in flow chart 1000 in Figure 10: STEP 1001: Form Data Matrix STEP 1002: Interrogate each element in the data matrix. STEP 1004: Identify all elements that are local maxima of intensity and have positive values. STEP 1006: Label each such local maximum as an ion. STEP 1008: Extract ion parameters. STEP 1010: Tabulate ion parameters. STEP 1012: Post-process ion parameters to obtain molecular properties. [00110] In Step 1008, the parameters of each ion are obtained by examining the maximal element. An ion's retention time is the time of the scan containing the maximal element. The ion's m/z is the m/z for the channel containing the maximal element. The ion's intensity is the intensity of the maximal element itself, or alternatively, the intensity can be the sum of intensities of elements surrounding the maximal element. Interpolation techniques, described below, can be used to better estimate these parameters. Secondary observable parameters, including for example, the widths of the peak in the chromatographic and spectral directions, can also be determined. Steps 2 and 3: Specification and Application of Filter Need for filters [00111] Rarely, if ever, is co-elution, interference, or noise absent in LC/MS experiments. The presence of co-elution, interference, or noise can severely reduce the ability to accurately and reliably detect ions. Consequently, the simple detection and quantification procedure illustrated by flow chart 1000 may not be adequate in all circumstances. Coelution [00112] Figure 11 is an exemplary contour plot showing the effects of co-elution and interference due to finite peak widths. In the example illustrated in Figure 11, another ion, ion 4, is assumed to have m/z and retention time values somewhat larger than that of ion 1 as well as have an apex in both the spectral and chromatographic directions lying within the FWHM of the apex of ion 1. As a result, ion 4 is co-eluted with ion 1 in the chromatographic direction and interferes with ion 1 in the spectral direction. [00113] Figure 12 illustrates the spectral effects due to co-elution of ion 4 at the times indicated by lines A, B, and C of Figure 11. In each spectrum shown in Figure 12, ion 4 appears as a shoulder to ion 1. This is also apparent from the contour plot shown in Figure 11 because there is no distinct apex associated with ion 4. [0100] Thus one problem with detection in LC/MS systems is that pairs of ions may co-elute in time and interfere spectrally such that the pair of ions produces only a single local maximum, not two. Co-elution or interference can cause true ions, having significant intensity in the data matrix, to be missed, i.e., not detected. Such missed detection of a true peak as an ion is referred to as a false negative. Noise [0101] Noise encountered in LC/MS systems typically falls into two categories: detection noise and chemical noise. Detector and chemical noise combine to establish a baseline noise background against which the detection and quantitation of ions is made. [0102] Detection noise, also known as shot or thermal noise, is inherent in all detection processes. For example, counting detectors, such as MCPs, add shot noise, and amplifiers, such as electrometers, add thermal or Johnson noise. The statistics of shot noise are generally described by Poisson statistics. The statistics of Johnson noise are generally described by Gaussian statistics. Such detection noise is inherent in the system and cannot be eliminated. [0103] The second kind of noise encountered in LC/MS systems is chemical noise. Chemical noise arises from several sources. For example, small molecules that are inadvertently caught up in the process of separation and ionization can give rise to chemical noise. Such molecules can be a constant presence, each producing an essentially constant background intensity at a given mass-to-charge ratio, or each such molecule can be separated thereby producing a chromatographic profile at a characteristic retention time. Another source of chemical noise is found in complex samples, which can contain both molecules whose concentrations vary over a wide dynamic range and interfering elements whose effects are more significant at lower concentrations. [0104] Figure 13 is an exemplary contour plot illustrating the effects of noise. In Figure 13,;numerically generated noise is added to an ion peak contour plot to simulate the effects of chemical and detector noise. Figure 14A illustrates mass spectra (Spectra A, B and C) corresponding to lines A, B and C respectively in Figure 13, Figure 14B illustrates chromatograms for ions 1, 2, 3 corresponding to lines labeled ion 1, ion 2 and ion 3 respectively in Figure 13. As can be seen in Figure 13, one detrimental effect of the additive noise is that it causes apices to appear throughout the plot, including within the FWHM of the nominal apex locations associated with ions 1 and 2. These noise-induced apices can be erroneously identified as peaks corresponding to ions, thereby resulting in false positive ion detections. [0105] Thus, local maxima may be due to the noise rather than ions. As a result, false peaks, i.e., peaks not associated with an ion, may be counted as an ion. Moreover, noise might produce more than one multiple local maximum for an ion. These multiple maxima could result in detection of peaks that do not represent true ions. Thus, peaks from a single ion could be multiply counted as separate ions when in fact the multiple peaks are due only to a single ion. Such detection of false peaks as ions is referred to as false positives. [0106] In addition to disregarding noise effects, the simple ion detection algorithm described in Figure 10 is generally not statistically optimal. This is because the variance in the estimates of retention time, m/z and intensity are determined by the noise properties of a single maximal element. The simplified algorithm does not make use of the other elements in the island of intensities surrounding the maximal element. As described in more detail below, such neighboring elements can be used to reduce variance in the estimate. Role of Convolution [0107] According to embodiments of the present invention, the LC/MS data matrix is a two-dimensional array. Such a data matrix can be processed by convolving it with a two-dimensional array of filter coefficients. [0108] The convolution operation employed in embodiments of the present invention provides a more general and powerful approach to peak detection than the simple signal-averaging schemes employed in conventional systems. The convolution operation employed in embodiments of the present invention addresses the limitations of the method described in Figure 10. [0109] The filter coefficients can be chosen to provide estimates of ion parameters that have better signal-to-noise ratios than those obtained from analyzing single channels or scans. [0110] The convolution filter coefficients can be chosen to produce estimates of ion parameters that have the greatest precision or least statistical variance for a particular data set. These benefits of embodiments of the present invention provide more reproducible results for ions at low concentration than do conventional systems. [0111] Another advantage of embodiments of the present invention is that filter coefficients can be chosen to resolve ions that are co-eluted and interfering. For example, the apices of ions appearing as shoulders to other ions in a mass spectrum can be detected using appropriately specified filter coefficients in embodiments of the present invention. Such detection overcomes limitations associated with conventional techniques in analyzing complex chromatograms, where co-elution and ion-interference are a common problem. [0112] Another advantage of embodiments of the present invention is that filter coefficients can be chosen to subtract baseline signals, producing more accurate estimates of ion intensity. [0113] Another advantage of embodiments of the present invention is that filter coefficients can be chosen to minimize the computation burden of convolution, resulting in high-speed operation of peak detection and the estimation of ion parameters. [0114] In general, numerous filter shapes can be used in the convolution, including, for example, Savitzky-Golay (SG) smoothing and differentiating filters. The filter shapes can be chosen to perform a number functions including smoothing, peak identification, noise reduction and baseline reduction. Filter shapes used in preferred embodiments of the present invention are described below Implementation of convolution in this invention [0115] The convolution operation according to embodiments of the present invention is linear, non-iterative and not dependent on the values of the data in the data matrix. In an embodiment of the present invention, the convolution operation is implemented by means of a general purpose programming language using a general purpose computer such as computer 118. In an alternate embodiment of the present invention, the convolution operation is implemented in a special purpose processor known as digital-signal-processor (DSP). Typically, DSP-based filtering provides enhanced processing speed over general purpose computer-based filtering. [0116] In general, convolution combines two inputs to produce one output. Embodiments of the present invention employ a two-dimensional convolution. One input to the two-dimensional convolution operation is the data matrix of intensities created from the spectral output of an LC/MS experiment. The second input to the two-dimensional convolution operation is a matrix of filter coefficients. The convolution operation outputs an output convolved matrix. Generally, the output convolved matrix has the same number of rows and column elements as the input LC/MS matrix. [0117] For simplicity in the present description, assume that the LC/MS data matrix is rectangular and that the size of the matrix of filter coefficients is comparable to the size of a peak. In this case the size of the filter coefficient matrix is smaller than the size of the input data matrix or output convolved matrix. [0118] An element of the output matrix is obtained from the input LC/MS data matrix as follows: the filter matrix is centered on an element in the input data matrix, and then the input data matrix elements are multiplied by the corresponding filter matrix elements and the products are summed, producing an element of the output convolved data matrix. By combining neighboring elements, convolution filters reduce variance in the estimates of an ion's retention time, mass-to-charge ratio, and intensity. [0119] The edge-values of the output convolved matrix are those elements that are within half the filter width from the edge of the output convolved matrix. Generally these elements can be set to an invalid value in embodiments of the present invention to indicate invalid filtering values. Generally, ignoring these edge values is not a significant limitation for embodiments of the present invention and these invalid values can be ignored in subsequent processing. One-dimensional Convolution [0120] Convolution for a one-dimensional case is clearly described in detail. This description is followed by generalizing convolution to the two-dimensional case. It is useful to first describe the one-dimensional case because the two-dimensional convolution operation that is used in the preferred embodiment of the present invention is implemented by applying a series of one-dimensional convolutions to the data matrix. [0121] In one dimension, the convolution operation is defined as follows. Given a one-dimensional, A/-element, input array of intensities dt and a one-dimensional, M-element, array of convolution filter coefficients f}, the convolution operation is defined as: (Equition remopved) where cr is the output convolved array, and i = 1,...,N. For convenience, M is chosen to be an odd number. The index j varies from j = -h,...,0,...h, where h is defined ash = (M -1)/2. [0122] Thus, the value of ci corresponds to a weighted sum of the h elements surrounding d,. Spectra and chromatograms are examples of one-dimensional input arrays that contain peaks. The width of the convolution filter /,. is set to be approximately the width of a peak. Thus, M is on the order of the number of array elements that span the width of a peak. Peaks have a width which typically is much smaller than the length N of the input array, so that in general M D N. [0123] Although the index /'for dt ranges from 1 to N, in some embodiments of the present invention, c, is defined only for i>hori< (N-h) to account for edge effects. The value for c, near the array boundaries, i.e. when / < h ori>(N-h),)s not defined for the summation. Such edge effects can be handled by limiting the values for c,. to be / > h or i < (N - h), where the summation is defined. In this case, the summation applies only to those peaks far enough away from the array edges so that the filter f] can be applied to all points within the neighborhood of the peak. That is, filtering is not performed at the edges of the data array dt. Generally, ignoring edge effects is no a significant limitation for embodiments of the present invention. [0124] If filtered values are needed near the edge for 1 < /' < h or N > i > (N - h), either the data array and/or the filter coefficients can be modified for these edge elements. The data matrix can be modified by append h elements to each end the array, and apply the M coefficient filter to an array that contains N + 2h elements. [0125] Alternatively, edge effects can be considered by appropriately modifying the limits of the filtering function to account for there being less than M points for filtering near the edges. Two-dimensional Convolution [0126] The one-dimensional convolution operation described above can be generalized to the case of two-dimensional data for use in embodiments of the present invention. In the two-dimensional case, one input to the convolution operation is a data matrix dtj subscripted by two indices, (i,j), wherein i = \,...,M and./ = l,...,Af. The data values of the input data matrix can vary from experiment to experiment. The other input to the convolution is a set of fixed filter coefficients, fpq, that is also subscripted by two indices. The filter coefficients matrix, fpq, is a matrix that has PxQcoefficients. Variables h and / are defined as h = {P~\)l2 and l=(Q-l)/2 . Thus, p = -h,...,h, andq = -l,...,I. [0127] Convolving dfj with fpq yields the output convolved matrix ciy.: (Equition remopved) [0128] Generally, the size of the filter is much less than the size of the data matrix, so that P«M and Q«N. The above equation indicates that ctJ is computed by centering f on the (i,j) th element of dtJ and then using the filter coefficients fpq to obtain the weighed sum of the surrounding intensities. Thus, each element of the output matrix cUj corresponds to a weighted sum of the elements of du, wherein each element dhJ is obtained from a region centered on the yth element. [0129] Although the index i andy for dLj ranges from i=1 to N, and j from 1 to M, in some embodiments of the present invention, ctj is defined only for i>hoxi<(N-h) and j>l orj<(M-/) to account for edge effects. Thevalue fore, near the array boundaries, i.e. when i(N-h) and/or j > I or j < (M -I) is not defined for the summation. Such edge effects can be handled by limiting the values for cUJ to be those where the summation is defined. In this case, the summation applies only to those peaks far enough away from the array edges so that the filter fpq can be applied to all points within the neighborhood of the peak. That is, filtering is not performed at the edges of the data arrays,. y. Generally, ignoring edge effects is no a significant limitation for embodiments of the present invention. [0130] If filtered values are needed near the edge for 1 < / < h and N > i > (N - h), either the data matrix and/or the filter coefficients marix can be modified for these edge elements. One approach is to append h elements to the end of each row, and / elements to the end of each column. The two-dimensional convolution filter is then applied to a data matrix that contains (N+2h)x(M + 21)elements. [0131] Alternatively, edge effects can be considered by appropriately modifying the limits of the filtering function to account for there being less than P points for filtering near the row edges and Q points for filtering near the column edges. [0132] The computational burden for implementation of equation (2) can be determined as follows. If fpq contains PxQ coefficients then the number of multiplications needed to compute a value for ctJ is PxQ. For example, where P = 20 and Q = 20, it follows that 400 multiplications are needed to determine each output point c,y in the output convolved matrix. This is a high computation burden that can be eased by other approaches to two-dimensional convolution. Two-dimensional Convolution with Rank-1 Filters [0133] The two-dimensional convolution filter described in equation (2) applies a filter matrix that contains PxQ independently specified coefficients. There are other ways for specifying the filter coefficients. Although the resulting convolution coefficients are not as freely specified, the computation burden is eased. [0134] One-such alternate way of specifying the filter coefficients is as a rank-1 filter. To describe a rank-1 convolution filter, consider that a two-dimensional convolution of the LC/MS data matrix can be accomplished by the successive application of two one-dimensional convolutions. See for example, in JOHN H. KARL, INTRODUCTION TO DIGITAL SIGNAL PROCESSING, PG. 320 (ACADEMIC PRESS 1989) ("KARL"), which is hereby incorporated by reference herein. For example, a one-dimensional filter, gq, is applied to each row of the LC/MS data matrix, producing an intermediate convolved matrix. To this intermediate convolved matrix, a second one-dimensional filter, fp, is applied to each column. Each one-dimensional filter can be specified with a different set of filter coefficients. Equation (3) illustrates how the filters comprising a rank-1 convolution filter are applied in succession, wherein the intermediate matrix is enclosed in the parentheses. (Equition remopved) [0135] The computational burden for implementation of equation (3) can be determined as follows. If fp contains P coefficients and gq contains Q coefficients, then the number of multiplications needed to compute a value for cu is P+Q. For example, where P = 20 and Q = 20, only 40 multiplications are needed to determine each output point ctJ in the output convolved matrix. As can be seen, this is more computationally efficient than the general case of two-dimensional convolution described in Eq. (2) where 20x20 = 400 are required to determine each cu. [0136] Equation (4) is a rearrangement of equation (3) that illustrates that the successive operations are equivalent to a convolution of the data matrix with a single coefficient matrix whose elements are pair-wise products of the one dimensional filters. An examination of equation (4) shows that in using the rank-1 formulation, the effective two-dimensional convolution matrix is a rank-1 matrix formed by the outer product of two one-dimensional vectors. Thus, equation (4) can be rewritten as: (Equition remopved) The two-dimensional coefficient matrix Fpq emerges from the convolution operation. F has the form of a rank-1 matrix, where a rank-1 matrix is defined as the outer product of a column vector (here, fp) and a row vector (here, gq). See for example, in GILBERT STRANG, INTRODUCTION TO APPLIED MATHEMATICS, 68FF (WELLESLEY-CAMBRIDGE PRESS 1986) ("STRANG"), which is hereby incorporated by reference herein. [0137] In embodiments of the present invention using a rank-1 filter implementation, the rank-1 filter is characterized by two orthogonal cross sections, one for each filter. The filter for each orthogonal cross-section is specified by a one-dimensional filter array. Two-dimensional Convolution with Rank-2 Filters [0138] A two-dimensional convolution operation can be carried out with a rank-2 filter. Two-dimensional convolution with a rank-2 filter is carried out by computing two rank-1 filters and summing their result. Thus, four filters: fp>Slq>fp>and s\ are required to implement a rank-2 filter for the two-dimensional convolution performed in embodiments of the present invention. [0139] Two of the filters fp and g1 are associated with the first rank-1 filter and two of the filters fp and g are associated with the second rank-1 filter. These four filters fp,fp and g\,g2q are implemented as follows: (Equition remopved) [0140] Filters fp and f2 are applied in the spectral direction (along the columns) and filters gq and g2q are applied in the chromatographic direction (along the rows). Equation (7) illustrates how each filter pair can be applied in succession, where the intermediate matrix is enclosed in the braces, and how the results from the two rank-1 filters are summed. Equation (7) shows the preferred manner of implementing the rank-2 filter according to embodiments of the present invention. [0141] Equation (8) is a rearrangement of equation (7) to show that the successive operations in the rank-2 filter configuration are equivalent to a convolution of the data matrix with a single coefficient matrix whose elements are the sum of pair-wise products of the two one-dimensional filter pairs. [0142] To analyze the computational requirements of a rank-2 filter, consider that if fp and fp both contain P coefficients and g\ and g\ both contain Q coefficients, then the number of multiplications needed to compute a value for an element of the output convolution matrix cUj is 2(P+Q). Thus, in the case where P = 20 and Q = 20, only 80 multiplications are needed to compute each element of the output convolution matrix, whereas in the general case as shown in equation (2), 20x20 = 400 are required to compute each chJ. [0143] Thus, an embodiment of the present invention employing a rank-2 filter, the effective two-dimensional convolution matrix is formed from the sum of the outer product of two pairs of one-dimensional vectors. Equation (8) can be rewritten as (Equition remopved) [0144] Two-dimensional coefficient matrix Fpq emerges from tne convolution operation. The two-dimensional coefficient matrixF has the form of a rank-2 matrix, where a rank-2 matrix is defined as the sum of two linearly independent rank-1 matrices as described in STRANG. Here fxpgq and fpgq are each rank-1 matrices. Filter Specifications [0145] Equations (2), (3), and (7) are all embodiments of two-dimensional convolution filters of the present invention. Equation (2) specifies the filter coefficients as a matrix / q, equation (3) specifies the filter coefficients as a set of two one-dimensional filters, fp andg?, and equation (7) specifies the filters as a set of four one-dimensional filters, fp, g\ and fp , g2q. [0146] Equations (2), (3), and (7) do not specify the preferred values of these coefficients. The values of the filter coefficients for the present invention are chosen to address the limitations of the method of Figure 10. The filter coefficients are chosen to accomplish several goals which include the reduction of the effects of detector and chemical noise, the partial resolution of coeluted and interfered peaks, the subtraction of baseline noise, and achievement of computational efficiency and high-speed operation. [0147] The Matched Filter Theorem (MFT) is a prescriptive method, known in the prior art, to obtain filter coefficients than can be implemented using Equation (2). See for example, KARL at 217; BRIAN D.O. ANDERSON & JOHN B. MOORE, OPTIMAL FILTERING 223ff (PRENTICE-HALL INC. 1979)("ANDERSON") at 223ff which is hereby incorporated by referral herein. Filters obtained from the MFT are designed to detect the presence of signals and to reduce the effects of detector noise. Such filters can then be used to detect ions in the LC/MS data matrix and can be used to determine the retention time, mass-to-charge ratio, and intensity of ions. A filter obtained from the MFT is an improvement over the method of Figure 10. In particular such filters reduce variance and improve precision by combining data from elements within a peak that neighbor the peak apex. However, such filters are not designed to subtract baseline noise or to resolve coeluted and interfered peaks. Filters obtained from the MFT and are not designed to achieve high speed operation. [0148] The MFT and a set of filter coefficients that can be obtained from it represent an improvement over the method of Figure 10 are described, then modified filters that subtract baselines, reduce the effects of coelution and interference, while still reducing the effect of detector and chemical noise are described. Such filters employ a combination of smoothing and second-derivative filters and are implemented using Equations (3) and (7). The preferred embodiment uses equation (7) with a combination of smoothing and second-derivative filters that together reduce noise, resolve interfering peaks, subtract baselines, and reduce the computational burden to allow for highspeed operation. Matched Filter Theorem for One-dimensional convolution [0149] The MFT is first described for one-dimensional convoution. It is then generalized to two-dimensional convolution. [0150] Coefficients for f} are chosen to perform a detection function. For example, the matched filter theorem (MFT) provides a set of filter coefficients known as a matched filter that can be used to perform the detection function. [0151] The MFT assumes that the data array dt can be modeled as a sum of a signal r0st plus additive noise,«.: The shape of the signal is fixed and described by a set of coefficients, st. The scale factor ro determines the signal amplitude. The MFT also assumes that the signal is bounded. That is, the signal is zero (or small enough to be ignored) outside some region. The signal is assumed to extend over M elements. For convenience, M is typically chosen to be odd and the center of the signal is located ats0. If H is defined as h = {M-\)/l , then s, = 0 for i < -h and for i > h. In the above expression, the center of the signal appears at i = h ■ [0152] For purposes of simplifying the present description the noise elements ni are assumed to be uncorrelated Gaussian deviates, with zero mean and a standard deviation of a0. More general formulations for the MFT accommodate correlated or colored noise. See example, ANDERSON at 288-304. [0153] Under these assumptions, the signal-to-noise ratio (SNR) of each element is r0sja0. The SNR of a weighted sum of the data that contains the signal ss can be determined by considering an M-element set of weights w,, centered to coincide with the signal where h = {M~\)/2 , and i = -h,...,0,...h. Assuming the weights are centered to coincide with the signal, the weighted sum S is defined as: (Equition remopved) [0154] The mean value of the noise term in an ensemble average is zero. Consequently, the average value of S over an ensemble of arrays, where the signal in each array is the same, but the noise is different is: (Equition remopved) [0155] To determine the noise contribution, the weights are applied to a region containing only noise. The ensemble mean of the sum is zero. The standard deviation of the weighted sum about the ensemble mean is: (Equition remopved) [0156] Finally, the SNR is determined as: (Equition remopved) This result is for a general set of weighting coefficients^.. [0157] The MFT specifies values for wt that maximize the SNR. If the weighting factors w, are regarded as elements of an M dimensional vector w of unit length, i.e., the weighting factors are normalized so thai (Equition remopved) = 1, then the SNR is maximized when the vector w points in the same direction as the vector s. The vectors point in the same direction when respective elements are proportional to each other i.e., when wi x Si. Consequently, the MFT implies that the weighted sum has the highest signal-to-noise when the weighting function is the shape of the signal itself. [0158] If W i is chosen such that wi = sr, then for noise with unit standard deviation, the SNR reduces to: (Equition remopved) This formulation of SNR corresponds to the signal properties of the weighted sum when the filter coefficients are centered on the signal and the noise properties when the filter is in a noise-only region. Matched Filter Theorem for Two-dimensional convolution [0159] The MFT discussed above for the one-dimensional case can also be generalized to the two-dimensional case for a bounded, two-dimensional signal embedded in a two-dimensional array of data. As before, the data is assumed to be modeled as a sum of signal plus noise: (Equition remopved) wherein the signal S;jis limited in extent and whose center is located at (i0,j0) with amplitude r0. Each noise element nt] is an independent Gaussian deviate of zero mean and standard deviation a0. [0160] To determine the SNR of a weighted sum of the data that contains the signal S!} consider a PxQ-element set of weights vt;.,, wherein (Equition remopved) weights are centered to coincide with the signal. The weighted sum S is: (Equition remopved) [0161] The average value of S over the ensemble is: (Equition remopved) he standard deviation of the noise is: (Equition remopved) and the signal-to-noise ratio is: (Equition remopved) [0162] As in the one-dimensional case described above, the SNR is maximized when the shape of the weighting function is proportional to the signal, that is when wtj