GC Image (LCxLC Edition) Users' Guide

Blob Analysis

Blob Metadata and Computed Attributes

LC Image maintains computed attributes and metadata about each blob. The computed attributes include maximum peak value, retention times of the peak, etc. The metadata includes:

• chemical name,
• group name,
• constellation name,
• whether the blob is to be included or excluded in reporting, and
• the internal standard associated with the blob,

None of the metadata values are set during blob detection; all metadata are set with information provided during analysis.

To manually set the metadata for a blob, it first must be selected. After one or more blobs are selected, a click of the right mouse-button will invoke a popup dialog box for editing the metadata, such as illustrated in Figure 1.

Figure 1: The Blob Metadata popup dialog displays and allows setting of metadata about the blob, including chemical name, group name, etc.

The Blob Metadata popup dialog for a single blob displays the blob's unique ID number and several computed attributes:

• ID - a unique identification number assigned by LC Image;
• Area - the number of pixels in the blob, i.e., its size (in number of datapoints) irrespective of its values;
• Volume - the sum of the values of the pixels in the blob, i.e., the total of its values or total response;
• Peak value - the largest value (or apex) of the pixels in the blob;
• Peak location - the coordinates of the peak apex in each dimension;
• Internal standard - the blob ID of the internal standard (0 if not set);
• Volume ratio - the ratio between the volume of the blob and the volume of the associated internal standard.

If a single blob is selected, the popup dialog displays the current values for the chemical name, chemical group name, and constellation name and allows those values to be changed. For a single blob, there are checkboxes to indicate whether or not the blob is to be included in reporting, whether or not the blob is an internal standard, and whether or not the blob is to be excluded in reporting. A blob can be included or excluded, or neither, but not both. A blob that is an internal standard must be included in reporting. There also are flags for creating text labels and/or chemical a structure annotation (as described in Text and Chemical Structure Annotations) based on the chemical name. Finally, there is a Set Color flag. If the flag is set for a blob, the default color (defined for the class) can be overridden with the color chooser.

If multiple blobs are selected, the popup dialog does not display the values specific to individual blobs. The multiple blobs dialog, illustrated in Figure 2, allows setting the chemical group name of all selected blobs, setting the constellation name of all selected blobs, clearing the names of all selected blobs, setting whether or not the selected blobs are to be included in reporting, setting whether or not the selected blobs are internal standards, setting whether or not the selected blobs are to be excluded in reporting, setting whether or not the blobs should have text labels, setting whether or not the blobs should have chemical structure, and setting a color for all selected blobs.

Figure 2: The Blob Group Metadata popup dialog allows setting of metadata about a selected group of blobs.

The internal standard of a blob that is not itself an internal standard is, by default, set to the nearest blob that is an internal standard. The defaults are applied when blobs are included and when an internal standard is included or excluded. The internal standard of an included blob can be changed by selecting the blob and cycling the association from one internal standard to the next until the correct association is made. Cycle the internal standard by first selecting an included blob (that is not an internal standard) and then clicking the Cycle Internal Standard from the Image Viewer tool bar or selecting Edit -> Cycle Internal Standard from the menu.

Clicking the Clear All Metadata button clears all blob metadata as well as all graphics and text objects. The user must confirm execution of this operation.

Blob Table

LC Image can present metadata and computed attributes for all blobs in an interactive table. Clicking the Show Blob Table button on the Image Viewer tool bar or selecting View -> Blob Table from the menu invokes a sortable table with a row for each detected blob. On each row is that blob's unique ID, metadata, and computed attributes. The Blob Table has the option to switch between three different tables, blobs, excluded blobs, and areas. The Blobs Table shows all blobs, except those that have been excluded. Excluded Blobs Table show only those that have been excluded. The Areas Table shows the data computed for all areas (graphics objects that are marked as areas, as described in Graphics and Areas, and analyzed similarly to blobs). Figure 3 illustrates a portion of an example blob table.

Figure 3: The Blob Table displays information about each blob. The table can be sorted by clicking on a column header.

LC Image supports more than 80 of computed attributes and metadata for blobs. The white paper, LCxLC Blob Metadata and Statistics in LC Image , describes in detail each of the available metadata and computed attributes in LC Image.

The blobs in the table can be sorted in forward or reverse order by clicking on the header of the column to be used as the sort key. (Consecutive clicks reverse the sort order.) Clicking to select a row (or using shift-click to select a range of rows or control-click to select multiple rows) causes the blob(s) selected in the Blob Table to also become selected and highlighted graphically in the Image Viewer. Right-click on a row brings up the metadata dialog for the indicated blob.

The number of decimal digits displayed for floating-point attributes can be set individually. To change the number of decimal digits, click the right button of the mouse on the column header to bring up the precision control dialog. Use the up or down arrows to increase or decrease the number of digits, then click elsewhere to set the indicated number of digits.

Some blob metadata also can be edited directly in the blob table. To edit a field, select the blob in the blob table and click on the field to be edited. Computed values cannot be edited. Metadata fields that can be edited display a text box when clicked. The Compound Name field provides a pull-down menu containing the hit-list returned by MS search (described in LCxLC-MS Data Processing and Analysis).

The blob table columns can be configured via the Configure -> Configure Blob Table on the Image Viewer menu. The Configure Blob Table dialog, pictured in Figure 4, allows the user to add or remove columns from the blob table configuration. Column(s) can be reordered by removing and re-inserting entries in a new position.

Figure 4: The Configure Blob Table allows the user to select the columns of the blob table.

Excluded blobs are not shown in the blob table. However, the blob table pane has a checkbox for toggling between the blob table and the table of excluded blobs.

The blob table can be exported as a text file from the File -> Save Blob Table menu item on the Image Viewer. The exported blob table will contain only the columns specified by the blob table configuration. The blob table is saved as text in comma-separated-values (CSV) format, which can be imported directly in the Microsoft Excel and other data analysis software.

Linear Retention Indices

Blob features for linear retention indices (LRI I and LRI II, for columns 1 and 2, respectively) are calculated based on retention index calibration tables. The retention index calibration tables are used to compute the linear retention index for all analytes (blobs and area objects) that have a peak (apex) retention time between that of the calibrants with the smallest and largest retention times. The linear retention index for an analyte is computed by linear interpolation between the two calibrants whose retention times most closely bracket the retention time of the analyte:

where T_a and I_a are the retention time and linear retention index of the analyte, T_n and I_n are the retention time and linear retention index of the calibrant with largest retention time less than T_a, and T_n+1 and I_n+1are the retention time and linear retention index of the calibrant with smallest retention time larger than T_a. For example, if an analyte has a first-column retention time of 35 min and the bracketing calibrants are octane and nonane with first-column retention times of 20 min and 40 min and retention indices 800 and 900, respectively, then the analyte's first-column linear retention index (LRI I) is: 875. The linear retention indices for analytes that have retention time less than that of the calibrant with the smallest retention time or larger than that of the calibrant with the largest retention time are not given a value by default.

The LRI Calibration Tables are accessed through the Configure menu.

1. Select Menu: Configure -> Configure LRI Table.
2. On the left hand side of the dialog is a list of two LRI tables. To switch between the tables simply select the desired table from the list.
3. To view the active LRI table as a graph, click the LRI Graph tab.

Figure 5: The Linear Retention Index configuration pane.

The software allows users to set LRI values at the limits of the chromatogram and supports extrapolation.

• Setting LRI values for start and end time: The LRI values for the limits are entered into the Retention Index column of the “Start Time Index” and “End Time Index” rows (These two names cannot be changed and are reserved.). The image start time and end time are loaded automatically from the image itself.
• Extrapolation: Currently, the only extrapolation settings are “No Extrapolation” or “Extend Lines”. These settings are used only in the event that a blob’s retention time is outside the calibration range of the LRI table.

Import an Existing LRI Table

LRI tables can be imported from external CSV files. The format is

(Chemical name), (Retention Time), (Index)

1. Select Menu: Configure -> Configure LRI Table.
2. Select the desired LRI table from the list on the left.
3. Click the Import button below the LRI table.
4. Use the file chooser to locate a CSV file and click Open.
5. Choose to Link or Copy the LRI table.

Link an External LRI Table

Linking a LRI calibration table to an image means that only the location of the file is stored within the image. The label for the file path, above the table, will be displayed as Linked To. A linked table is not editable within LC Image. When the external file is changed, LC Image will automatically refresh the table when opening the image or a user can manually refresh the table through the Refresh button: .

Copy an External LRI Table

Copying a LRI calibration table to an image means that all records in the external file are copied to the image. A copied table is editable within LC Image. The label for the file path, above the table, will be displayed as Copied From. When the external file is changed, the table will not be affected.

Edit the LRI Table

If the LRI table is imported from an existing table, then editing is allowed only if the Copy option was chosen. When editing the table, all retention times and indexes must maintain a strictly increasing order.

• Add a Row to the LRI Table:
  Add Row adds a blank row to the active LRI table. The row will sorted automatically based on the retention time.
• Delete a Row from the LRI Table:
  Delete Row deletes the selected row(s) from the active LRI table.
• Clear LRI Table:
  Clear removes all rows from the active table.

Populate LRI Table

LC Image can automatically generate a new table and populate the Chemical Name and Retention Time columns from the named and included blobs in the image. This action will overwrite any data currently in the table.

1. Select Menu: Configure -> Configure LRI Table.
2. Select the desired LRI table from the list on the left.
3. Click the Populate button from the row of buttons below the LRI table.
4. Fill in the index values to be used for each compound.

Refresh a Linked LRI Table

LC Image automatically refreshes a linked LRI table from the linked file when opening an image. A user also can refresh a linked table manually.

1. Select Menu: Configure -> Configure LRI Table.
2. Select the desired LRI table from the list on the left.
3. Click Refresh:

Update LRI Table from Image

LC Image can automatically update the retention times in a LRI table with the retention times of the named blobs from an image. In order to update, the table must have chemical names that match blobs from the image and the table must not be a linked table.

1. Select Menu: Configure -> Configure LRI Table.
2. Select the desired LRI table from the list on the left.
3. Click the Update button from the row of buttons below the LRI table.
4. A dialog will display which Compounds / Rows were updated or a message indicating that no rows were updated.

Export LRI Table

A LRI table can be saved to an external CSV file.

1. Select Menu: Configure -> Configure LRI Table.
2. Select the desired LRI table from the list on the left.
3. Click the Export button from the row of buttons below the LRI table.
4. Choose a location to save the table, then click Save.

Reference Peak

A Reference Peak can be assigned to a blob to provide additional statistics for quality control, including retention-time differences, resolutions, and volume ratio. Reference Peaks can be set for either blobs or template peaks. (Templates are described in Chemical Identification.) Blob reference peaks can reference only other un-excluded blobs and template reference peaks can reference only other template peaks.

• Set Reference Peak in Blob Properties Dialog
  1. Select Blob Mode.
  2. Select a Blob and right click.
  3. Select the “Analysis” tab.
  4. Type in a Blob ID in the “Reference Peak” field or select one of the options listed in the drop-down menu, as in Figure 6.
  5. Click OK to apply.

  Figure 6: The Blob Metadata popup analysis tab allows for setting the reference peak.

• Reference Peak Selection Options
  The drop-down menu for the “Reference Peak” allows the user to browse for the reference peak, choose the current internal standard blob, or choose the current nearest blob.  If internal standard or nearest blob is chosen the reference is set to that particular blob and will not be updated, should internal standard or nearest blob change.
  • Assign Reference Peak by Entry
    Enter the Blob ID for the desired reference peak.  If the Blob ID entered is not valid, a warning message will be displayed to inform the user that the ID could not be found.
  • Assign Reference Peak using Browse
    Choose “Browse...” from the drop-down menu to open a mini-blob table.  This table shows the Blob ID, Compound Name and Peak location for each un-excluded blob, as illustrated in Figure 7.  To choose a reference peak, select the corresponding row then press “OK”. Double clicking on a row in the table will automatically choose that row and close the table.  The Mini-Blob Table also can filter blobs based on compound name.

    Figure 7: Browsing for reference peak

  • Assign Reference Peak to Internal Standard
    If the current blob has an internal standard set, then this option will appear in the drop down list. Choosing this option will put the current internal standard blob's ID in the reference peak field.
  • Assign Reference Peak to Nearest Blob
    Choosing this option will put the current nearest blob's ID in the reference peak field.
• Blob Features for Reference Peak
  There are several computed attributes related to reference peak. They are:
  1. Reference Peak — The Blob ID of the reference peak.
  2. Reference Compound — The Compound Name of the blob indicated as the reference peak.
  3. Reference Difference I — The first column difference (in time or samples, as configured by the user) between the apex of the blob and the apex of the reference blob.
  4. Reference Difference II — The second column difference (in time or samples, as configured by the user) between the apex of the blob and the apex of the reference blob.
  5. Reference Volume Ratio — The volume ratio between the blob and the reference blob.
  6. Reference Volume-to-Concentration Ratio — The volume-to-concentration ratio between the blob and the reference blob (requires that the amount table have concentrations for both compounds).
  7. Reference Resolution attributes include Resolution 2D, Reference Resolution I, Reference Resolution II, and Reference Resolution 2D.

Global CLICs

LC Image allows users to perform customized global qualitative/quantitative analysis on all blobs with Global CLIC in Blob Table. Global CLIC supports blob analysis using the Computer Language to Identify Chemicals (CLIC) to create powerful formulas. The CLIC Tool is described in Chemical Identification. The result of that CLIC expression is shown in a new column in the blob table. Users can use the functions available in CLIC expressions to build Global CLICs. Any CLIC expressions that were saved through the CLIC tool for selecting blobs or extracting CLIC images are available to Global CLICs.

The result of a Global CLIC is always a number. If the CLIC expression entered results in true for a given blob, the result is shown as 1. False results are shown as 0. Any results that are strings are shown as NaN (Not a Number).

• Create Global CLIC
  1. Open Blob Table.
  2. Click on the Create CLIC column button:
  3. Enter the desired CLIC expression or select one from the selection box and click OK.
• Edit Global CLIC
  1. Right click on the column header for that CLIC column.
  2. Select Edit CLIC.

     

• Delete Global CLIC
  1. Right click on the column header for that CLIC column.
  2. Select Delete CLIC.
• Delete All Global CLICs
  1. Open Blob Table.
  2. Click on the Remove all CLICs button:
  3. Click Yes.

Blob Sets

Blobs sharing the same group name form a blob set. Sets of blobs also are indicated by included graphic objects (points, polylines, rectangles, and polygons). As described in Graphics and Areas, graphical objects can be flagged to be included in reporting. Each graphical object that is flagged for inclusion defines a blob set. Point graphics define the set of blobs that contain the point. Polyline graphics define the set of blobs that are intersected. Rectangle and polygon graphics can be used to indicate the set of blobs whose peaks are intersected or contained in the graphic. A graphical object also may have a CLIC expression (described in Chemical Identification ) that imposes rule-based constraints on set inclusion.

LC Image can present information about each blob set in tabular form. Clicking the Show Blob Set Table button on the Image Viewer tool bar or selecting View -> Blob Set Table from the menu invokes a sortable table with a row for each blob set. Figure 8 shows an example blob set table.

Figure 8: The Blob Set Table displays information about blob sets. The table can be sorted by clicking on a column header.

On each row is the blob set's ID (for sets defined by graphics objects), name of the group or defining graphic, number of blobs in the set, number of included blobs in the set, total volume of all blobs in the set, total volume of all included blobs in the set, total fractional response (as percent) of all blobs in the set, and total fractional response (as percent) of all included blobs in the set.

Each row of the blob set table also has a color selection button. If the color is applied to a blob set, it overrides the color defined for the class (included, excluded, or unflagged) of blobs in the set, but not the colors assigned for specific blobs. The priority for colorizing blobs (from high to low) is blob, group, constellation, graphic(s), and class. If a blob is colorized by its graphic(s) and is contained in multiple sets defined by graphics that apply colors to blobs, then the blob is colored by the average color of those graphics.

The entries in the table can be sorted in forward or reverse order by clicking on the header of the column to be used as the sort key. (Consecutive clicks reverse the sort order.) Clicking to select a row (or using shift-click to select a range of rows or control-click to select multiple rows) causes the blob(s) in selected blob set(s) to become selected and highlighted graphically in the Image Viewer . The number of decimal digits displayed for floating-point attributes can be set individually by clicking the right button of the mouse on a column header, as described for Blob Tables.

The blob set table can be exported as a text file by selecting File -> Save Blob Set Table from the Image Viewer menu. The exported blob set table is saved as text in comma-separated-values (CSV) format, which can be imported directly in the Microsoft Excel and other data analysis software.

Blob Set Table Column Configuration

The blob set table columns can be configured via the Configure Blob Table menu. The Blob Set Table Configuration dialog, pictured in Figure 9, allows the user to control both:

1. Which columns are included in Blob Set Table and related reports;
2. The order in which the included columns are listed.

Figure 9: The Blob Set Table Configuration dialog.

To configure the Blob Set Table,

1. From the menu, select: Configure -> Configure Blob Set Table.
2. Add or remove columns from the visible feature list. Column(s) can be reordered by removing and re-inserting entries in a new position or using the promote and demote buttons.
3. Click OK to Apply.

The Blob Set Table configuration is saved with the image.  A specific configuration can be saved as the default configuration so that it persists across sessions.

Blob Set Table and Area Table Residuals

The software includes a row in the Blob Set Table that reports the computed attributes for all blobs not included in any blob set, as in Figure 10. The Areas Table has a similar residual row that reports the computed attributes for all regions not included in any area, as in Figure 11.

Figure 10: Residual Set in Blob Set Table.

Figure 11: Residual Area in Area Table.

Blob Selection Tools

LC Image provides three other tools for blob selection: Find Blobs, Analyze Blobs, and CLIC Tool.

The Find Blobs tool, pictured in Figure 12, supports blob selection on the basis of blob ID, compound name, group name, constellation name, and chemical formula. The search tool also supports wildcards. For example, all blobs with a specific group name could be selected en masse. This tool is accessible from the Edit menu.

Figure 12: The Find Blobs tool allows selection of blobs on based metadata.

The Analyze Blobs is described in the next section. The CLIC Tool supports blob selection using the Computer Language to Identify Chemicals (CLIC) to create powerful rule-based constraints. The CLIC Tool is described in Chemical Identification.

Analyze Blobs

LC Image provides a Blob Analysis Tool, pictured in Figure 13, for graphing bi-variate relationships between blob statistics and for identifying blobs with particular statistical attributes.

Figure 13: The Blob Analysis Tool graphs bi-variate relationships of statistics.

The Blob Analysis Tool also is useful for identifying blobs that share certain statistical characteristics. For example, Figure 13 shows that the relationship between Peak Value and Volume is roughly linear for this set of blobs.

The main operations of the Blob Analysis Tool are:

• Graphing
• Edit Blob Metadata

To open the Analyze Blobs tool, blobs must be detected. If no blobs are selected when the tool is opened, then all blobs will be used for the analysis. If a subset of blobs is desired, select them before opening the tool.

1. Select a set of blobs in Blob mode.
2. Select Menu: View -> Analyze Blobs.

Graphing

Graphing with the blob analysis tool is accomplished by selecting a feature for the x and y axes. The graph will be redrawn, plotting each blob as a small circle. If a blob has a set color, then that color will be used to draw that blob's circle. Graphing features for non-MS images are limited to the set of numeric blob features. For MS images, blobs can be graphed based on MS channels or on PCA for the blobs' MS channels.

• Data Pre-Processing can be used to normalize spectra in MS images for Multi-Spectrum Range and PCA analysis, based on the intensity of the largest ion peak. Check the Normalize Multi-Spectrum checkbox, to enable this option. If the current graph is using Multi-Spectrum Range or PCA and the Normalize checkbox is selected or deselected the graph will automatically be updated.
• Blob Features of any numeric blob features can be used for either axis. The list of blob features that is shown in the combo boxes are the same as the list of features visible for the Blob Table. So, if the desired blob feature isn't shown in the combo box, go to the Configure Blob Table menu item and add the feature to the visible list. For non-MS images, the only graphing capabilities are for blob features, as shown in Figure 13.
• Multi-Spectrum Range allows the graphing of a specific spectral channel or range. Ranges are entered as commas separated lists (e.g., 104-110.2,125.75). The summed intensity of this range then is plotted for each blob.
• Principal Component lists the top ten principal components of the blobs Multi-Spectrum, e.g., in Figure 14. The loadings matrix shows the actual values found from the principal component analysis.

Figure 14: PCA used to separate blobs into groups (Red and Yellow).

Edit Blob Metadata

A user can select a group of blobs on characteristics associated with the x and y axes and then edit their metadata.

Edit Multiple Blobs:

Selecting multiple blobs requires the use of the polyline selection, as in Figure 15. Draw a polyline around the blobs to be edited. This line is persistent as long as the graph isn't changed. To clear the line, right click and choose Clear Selection. If the line is closed and needs to be extended choose Extend Current Selection. This will allow the polyline to be extended. After a polyline is used to select a set of blobs, the Retrieve Blobs button will load only the selected blobs to the graph. Then, PCA is recomputed on the new set of blobs and a new Loadings Matrix is generated.

1. Left click the mouse on the graph to begin drawing selection line.
2. Continue left clicking as vertices are created to enclose blob group.
3. To close selection line double click
4. Right click on graph to display right click menu

   

5. Choose Edit Properties to change the metadata for the blobs inside selection polygon.
   Or click the Edit Properties button.

Figure 15: The Blob Analysis Tool with designated blobs enclosed by a polygon.

Edit Single Blob:

To edit a single blob, left click on the blob. The circle of the blob that is selected is then filled, indicating its selection. The Blob Properties dialog will pop up automatically. If the mouse is placed over a particular blob, a tooltip will display the blob mouse-over. The information shown in the tooltip is configured through the Configure -> Configure Setting -> Blob Mouse-Overs.

Cluster Blobs

LC Image can separate selected blobs into clusters based on their retention-time proximity — that is, the selected blobs are partitioned so that nearby blobs tend to be placed in the same partition and distant blobs tend to be placed in different partitions. For each cluster, LC Image creates a convex polygon enclosing the inscribing retention-time rectangle of each blob in the cluster. Depending on the configuration and parameterization, the cluster polygons may be included (or not) or added to the template (or not). (Templates are described in Chemical Identification.)

To Cluster Blobs, enter Blob mode, select the blobs to be clustered, and select "Edit -> Blob Clustering. Then, in the Cluster Blobs dialog, pictured in Figure 16, choose the clustering algorithm, the scaling of the first-column retention (in pixels) relative to the second-column retention (in pixels), the number of clusters, whether or not to include the cluster polygons, and whether or not to add the cluster polygons to the template.

Figure 16: The Cluster Blobs dialog.

Figure 17 pictures four cluster polygons partitioning selected blobs. If the cluster polygons are included, then the blob set for the cluster will be included in the Blob Set Table. The cluster polygons also can be added automatically to the template. As described in Chemical Identification if the cluster polygons are added to the template, marker peaks within each cluster also may be determined automatically and added to the template.

Figure 17: Four cluster polygons drawn to partition selected blobs.

Here are the brief descriptions on each clustering algorithm. All blobs are represented by peak locations in pixel coordinates.

• K-means: K-means clusters blobs by minimizing the sum-of-squared Euclidean distances from each blob to its cluster center. It is suitable for recovering compact clusters.
• Hierarchical clustering algorithms: The hierarchical clustering algorithms, cluster blobs by merging clusters one pair at a time. They start with clusters each with a single blob and keep merging the two closest clusters until the required number of clusters exists. Based on different measures of proximity, there are four algorithms:
  • Single linkage: The proximity of two clusters is the minimum distance between two blobs in the two clusters. It tends to generate elongated clusters.
  • Complete linkage: The proximity of two clusters is the maximum distance between two blobs in the two clusters. It tends to generate compact clusters.
  • Average group linkage: The proximity of two clusters is the average distance between two blobs in the two clusters. It is a compromise between single linkage and complete linkage.
  • Complete linkage with PCA: The proximity of two clusters is the determinant of covariance matrix of the union of the two clusters in their PC (principal component) space, whose square root is proportional to the area or volume generated by the data. It tends to find natural clusters.

Contents Previous: Peak Detection Next: Graphics and Areas

GC Image (LCxLC Edition)™ Users' Guide © 2001–2011 by GC Image, LLC, and the University of Nebraska.