Chromatographic fingerprinting by template matching is significant because it can automatically extract rich untargeted and targeted chemical information by GC times GC chromatograms across multiple samples with a consistent cross alignment. The main advantages of this technique are its comprehensive coverage of untargeted and targeted analytes, and its consistent matching of chromatographic features across multiple chromatograms and samples by cross alignment. The technique has broad applications.
Here we apply it to foodomics and specifically to characterize the chemical information encrypted in the volatile fraction of extra virgin olive oils. The technique is foundational for research using machine learning, to discover complex relationships between sample chemistry and external factors. For example in food, using sensory quality and production methods.
I would call it the chromatography is essential. If the separations are poor, the power of the technique is limited, but if the chromatograms are good, the method is a fairly successful one GC by GC visualizes structure retention relationships in images. The human eye is well adapted to perceive two dimensional patterns, and so seeing the method demonstrates its power in a very direct manner.
To create a two dimensional raster array for visualization and processing, launch the image software and select file and import to open the raw data files of interest. In the import dialogue set the modulation period to 3.5 seconds, click OK, and select file and save image as. Navigate to the desired folder, enter the file name, and click save.
To shift the modulation phase select processing and shift phase and set the shift amount to minus 0.8 seconds. For baseline correction, select graphic and draw rectangle and click and drag to draw a rectangle in the image in which no peaks are detected. Select tools and visualize data and note the mean and standard deviation of the detector signal.
Then close the tool and select processing and correct baseline. To color the chromatographic images using value and color maps, select view and colorize. Open the import export tab and select the number AAAA custom color map, and click import.
Then in the value mapping controls, set the value range to the minimum and maximum values and click okay. For 2D peak anolyte detection, select processing and detect blobs. Some peaks will be split and some spurious detections will be observed.
Select configure, settings and blob detection. Set smoothing to 0.1 for 1D samples and to two for 2D modulations, and set the minimum to one times 10 to the sixth. Then click okay, select processing, and detect blobs and observe the improvements.
To remove meaningless detections, first select processing and interactive blob detection. Note the blob detection settings, and click detect. In the advanced filter builder, click add.
In the new constraint window select Retention 2, and click okay. Then in a constraint sliders set the minimum and maximum 2D retention times for the filter to reduce the number of false peaks without losing the true peaks and click apply and yes. To search for the peak spectra and a NIST MS library of interest select configure, settings and search library and set the type of spectrum to peak MS, the intensity threshold to 100, the NIST search type to simple similarity, the NIST retention indices column type to standard polar, and the NIST retention indices tolerance to 10.
Then click OK and select processing and search library for all blobs. To create a template with targeted peaks in the image view, click on the first peak and control click on any additional peaks to select the peaks of interest. Click add to template and select file and save template.
Then specify the folder and file name and click save. To match and apply the template, select file and open image and navigate to and open the chromatogram file of interest. Set the cursor mode to template and select objects and select template and load template.
In the load template window, click browse and open the targeted peaks template. Then click load and dismiss. In the image view, right click on a template peak to inspect its object properties, including the qualifier chemical logic expression, or qCLIC, and reference MS.Select template and interactive match and transform template.
In the match template window, click match all and review the matching results in the table and in the image. Each template peak will be marked with an unfilled circle. If a match is made, there will be a link to a filled circle for the detected peak.
Then edit the matches as appropriate and click apply to transfer the metadata from the template to the chromatogram. If chromatographic conditions vary substantially, causing the template to be misaligned with a new chromatogram, load the targeted template2. bt file and select template and interactive match template and click edit transform.
In the transformed template interface vary the 1D and 2D scales, translations, and shears to better align the template with the detected peaks, and click transform template. When the template has been transformed, click edit match, and match the template peaks as demonstrated. To establish correspondences between untargeted and targeted analytes select file and open analysis and open the file as indicated.
Open the compounds tab to review the metric values and statistics for specific analytes or untargeted analytes with identifiers aligned across all of the chromatograms. Open the attributes tab to review values and statistics for specific metrics across the chromatograms and open this statistical summary tab to review the summary statistics for both compounds and features. If the chromatograms are from different classes, the summary tab will list Fisher Ratio statistics that provide insight into the features for discriminating between classes.
In these graphs, relative retention patterns for homologous series and classes are shown. With annotations for linear saturated hydrocarbons, unsaturated hydrocarbons, linear saturated aldehydes, monounsaturated aldehydes, polyunsaturated aldehydes, primary alcohols and short chain fatty acids. Detected 2D peaks can be identified by comparing the average MS spectrum extracted from the entire 2D peak or from the largest spectrum.
The collection of identified 2D peaks can be adopted to build a template of targeted peaks to properly establish reliable correspondences between the same compound across all of the sample chromatograms. In this analysis, a partial misalignment between the targeted template and the actual chromatogram can be observed. For minimal misalignments, interactive template transformations can reposition the template peaks for a better fit.
The untargeted feature template is composed of 2D peaks from analytes detected in the composite chromatogram that are matched by the reliable peaks template. The mass spectra of the composite peaks, as well as their retention times are then recorded in the feature template. When unsupervised pattern recognition by principal component analysis is applied to targeted peaks distribution within the 20 analyzed samples, Sicilian and Tuscany oils cluster separately, suggesting that pedo-climactic conditions and terroir impact the relative prevalence of volatiles.
When first learning the procedure it is important to check the results at each step. If you have any questions, the people at GC Image are very responsive and helpful. We are currently using the technique to analyze other food products, such as for example, hazelnuts, cocoa, and teas, but also bio-fluids, such as saliva, urine, and feces for nutrient metabolomic applications.
