Visualization and Multivariate Statistics

This video will demonstrate visualization and multivariate statistics steps in GC Image.

Note

Demos 4, 5, and 6 should be completed in sequence.

The public dataset used in this demo can be downloaded here:

Supporting files to help in following along with this quick start demo can be downloaded here:

A premade analysis file that can be used in this demo is included in the Chocolates Demo Processing Files linked above (Chocolate.gca/).

1 Check Chromatograms

We can visually compare the chromatograms using the Images perspective.

  1. From the Investigator/Analysis software, select File > Open Analysis.
  2. Browse to the save location of your previously saved analysis. Click Open.
  3. Switch to the Images perspective by clicking on the Images tab.
  4. Select the Chromatogram tab to view the chromatogram images.
  5. Click on items from the Explore Images list on the left of the window to view the image. This list shows the image name and class.

2 Pairwise Differences Plot

We can perform a more detailed pairwise comparison between classes using the pairwise differences plot.

  1. From the menu, select Analysis > Pairwise Differences.
  2. Switch to the Summary perspective by clicking on the Summary tab.
  3. Select the Bubble Chart tab and Areas from the list on the left.
  4. From the drop-down menus above the chart, select the following attributes to plot:
    • X: Retention I.Mean
    • Y: Retention II.Mean
    • Z: Percent Response.Mean Difference(MintLime, Orange) The z axis corresponds to size of the bubble and, in this case, indicates the magnitude of the difference of that feature between the classes.
  5. Select the largest bubble (around RT1 5.0, RT2 1.2) by clicking several times around the bubble to enclose it in a polygon.
  6. The table below will highlight rows based on your selection. In this case, the selected bubble is area (10).
  7. Switch to the Compounds perspective by clicking on the Compounds tab.
  8. Select Areas from the View drop-down menu, enter 10 in the Filter field at the bottom on the list on the left, and select area (10) from the list.
  9. Select the Distribution tab and Percent Response from the Y:1 drop-down menu above the chart to compare response values of area (10) across all samples
    • The responses for the MintLime and Orange classes are higher than those of the other classes. The Orange response is roughly twice that of MintLime.
  10. Select the Spectra tab to view and compare the spectra of area (10) across all samples.
    • The MintLime and Orange spectra are similar to each other but different from all other spectra.
  11. Click on one of the MintLime spectra and the Send to Library Search button to perform a NIST library search.
    • The library search identifies the most likely match for the compound as D-Limonene.

3 Compound Finder

The Compound Finder tool can identify common and unique compounds between sample classes based on spectral and response criteria.

  1. From the menu, select Tools > Compound Finder.
  2. On the Compound Finder settings window:
    • Click Defaults
    • Uncheck all three options under Cross Sample Significance Check
    • Adjust the Spectral Match Factor Threshold value to 700
    • Click OK
  3. The Compound Finder - Results perspective shows a bubble plot depicting the results:
    • The bubbles' x and y location correspond to retention time
    • The bubbles' size corresponds to percent response be default but can be changed with the Z: drop-down menu
    • The bubbles' color corresponds to either the sample class it is unique to or Common if it is present in all classes. The legend at the bottom of the chart indicates the color mapping.

3.1 Investigate a Common Compound

  1. Select the largest Common bubble (around RT1 7.0, RT2 1.8) by clicking several times around the bubble to enclose it in a polygon.
  2. The table below will highlight the row with area (17).
  3. Switch to the Compounds perspective by clicking on the Compounds tab.
  4. Select Areas from the View drop-down menu, enter 17 in the Filter field at the bottom on the list on the left, and select area (17) from the list.
  5. Select the Distribution tab and Percent Response from the Y:1 drop-down menu above the chart to compare response values of area (17) across all samples
    • The response is significant across all sample classes.
  6. Select the Spectra tab to view and compare the spectra of area (17) across all samples.
    • The spectra are similar across all sample classes.
  7. Click on one of the spectra and the Send to Library Search button to perform a NIST library search.
    • The library search identifies the most likely match for the compound as Pyrazine, tetramethyl- (Synonym: Tetramethylpyrazine).

3.2 Investigate a Unique Compound

  1. Return to the Compound Finder - Results perspective.
  2. Select a large MintLime bubble around RT1 8.0, RT2 1.5 by clicking several times around the bubble to enclose it in a polygon.
  3. The table below will highlight the row with area (69).
  4. Switch to the Compounds perspective by clicking on the Compounds tab.
  5. Select Areas from the View drop-down menu, enter 69 in the Filter field at the bottom on the list on the left, and select area (69) from the list.
  6. Select the Distribution tab and Percent Response from the Y:1 drop-down menu above the chart to compare response values of area (69) across all samples
    • The response is significantly higher for the samples in the MintLime class.
  7. Select the Spectra tab to view and compare the spectra of area (69) across all samples.
    • The MintLime spectra are similar to each other but different from all other spectra.
  8. Click on one of the spectra and the Send to Library Search button to perform a NIST library search.
    • The library search identifies the most likely match for the compound as Cyclohexanone, 5-methyl-2-(1-methylethyl)-, cis- (Synonym: Isomenthone).