3.65 MB

| Deutsch | | Español |

Doing Quantitative Analysis

For getting a qualitative impression of differences between two images only, simply open the Dual View for these images.

Finding Interesting Spots in a Scatter Plot

For finding interesting spots you can compare their expression levels in two images. In the spots menu of the Dual View please click on 'Spots -> Show Scatter Plot' to open a graphical representation of the spots of these two gel images.

Scatterplot

The position of a spot is determined by its normalized volume on each image: in this example, the x-position by its quantity in Control_01 and the y-position by its quantity in the 1min_01 image. Thus, a spot having an unchanged volume on both images appears on the 45 degree line of this graph, whereas induced spots are found in the upper left and reduced spots in the lower right part of this graph. Now just click on one of the spots in the top or bottom of the Scatter Plot and keep an eye on the Dual View. The selected spot shape will be highlighted in the Dual View, and the view will scroll to the selected spot if necessary.

Identifying Interesting Spots by Expression Profile

You can view an expression profile of a spot over the complete project: in the menu of the Dual View, please click on 'Rollups -> Expression Profiles'. A new window will open, keeping its position in front of its parent window: one of the so-called 'Rollups'.

Make sure that the Spot Selection Tool is selected and simply move the mouse pointer to a certain spot.

Expression profile rollup displaying the spot intensity throughout the whole project

The rollup dynamically shows the expression profile for this spot across the whole project. The height of each bar is determined by the relative volume (% Vol) of the spot, i.e. the spot's intensity after background subtraction and normalization. Now move the mouse pointer to the next spot and watch how the expression profile changes.

Filtering for Interesting Expression Profiles

Switch to the Quantitation Table by choosing 'Window -> Quantitation Table'. The quantitation table window shows spot data in three different views:

• single gel view
  Shows spot data such as relative volume, area, and ID for spots on a single gel image. The spot's quantity is in the % Vol column, it is the result of background subtraction, quantitation and normalization. You can open single gel image tables by clicking on the tab with a gel image's name at the bottom of the window.
• 'All gel images' view
  Shows spot data for all images in the project showing one expression profile in every row in the table. In addition to the %Vol (relative volume) columns this table also includes Ratio columns displaying expression ratios as color codes as well as numbers. All ratio columns are calculated in accordance to a common ratio master image, to be changed in the table properties (choose 'Columns -> Table properties'), if necessary.

Quantitation Table: 'Statistics View'

• 'Statistics' view
  Shows relative volumes as well as averages and relative standard deviations for groups, accompanied by t-Test results with respect to the first group. You can also compute ratios such as 'mean of group 1min / mean of group control'.

The Quantitation Table is synchronized with the other views. Selecting an expression profile will select the spot for example in the dual view or in the scatter plots. Delta2D offers powerful tools to identify relevant expression profiles in accordance to your criteria.

Filter Dialog

Let us try a simple example: We are looking for spots where intensities in both sample groups are increased or decreased by a factor of at least 2 relative to the control. This factor of 2 should apply to the means of the replicate groups, i.e. the mean intensity within group '1min' and '10min' should be at least two times greater or smaller than the mean intensity of group 'control'.

To see only the spots matching these criteria, follow the following steps:

1. Switch to the Quantitation Table and select the 'Statistics' tab.
2. To see the name of a column either drag the line between two column headers to make it wider or simply point to its header and wait until its name appears as a tool tip. The name of the column we are looking for is 'Ratio mean % volume 1min / mean % volume control'. Click on the top part of the column header labeled 'Filter' to open the respective filter.
3. Insert the filter borders '0.5' and '2' into the fields 'Show values from ... to ...'. Alternatively, you can drag the left slider below the histogram. The box 'Filter active' will be checked automatically, additionally check the box 'Negated'. The histogram highlights the distribution of the ratio values.
4. Press OK to close the dialog. The quantitation table will now show only those expression profiles including induced or reduced spots of a factor of at least 2.

Repeat the steps 2 - 4 above for the column 'Ratio mean % volume 10min / mean % vol-ume control'.
Open the Dual View again with the two images 'control_01' and '1min_01'. Only spots matching our criteria will be shown there.

Advanced Statistical Analysis

Since version 3.6 Delta2D includes advanced multivariate statistics for the analysis of 2D gels, including:

• Heat map display of expression profiles
• Various methods of clustering
• t-test variations
• Analysis of Variance (ANOVA)
• Template matching for expression profiles
• Principal Component Analysis (PCA)

Statistical analysis in Delta2D is based on the TIGR MeV (Multiple Experiment Viewer) and tightly integrated into the image analysis workflow. With Delta2D's 100 Percent Spot Matching, there are no missing values. Matching problems are virtually eliminated, making it especially suitable for the methods that were originally developed for DNA microarray analysis.

Most statistical algorithms are recommended for being applied to a minimum number of data sets while others even demand for such a minimum. The example project is too small to rely on results but still helps to understand how statistical analysis in Delta2D works.

Open the quantitation table ('Windows -> Quantitation Table') and make sure the Statistics Table is selected. Hide the quantitative data for the fused image: Choose 'Column -> Column Properties', uncheck the checkbox next to Fused Image. Then press OK and the Fused Image is excluded from the analysis.

Getting a high level overview of expression data - heat map

Press the Analyze button in the top left of the statistics table. A new analysis window is opened, containing the current expression profiles in a heat map display.

The legend across the top shows the color code for spot intensities. Rows are labeled based on the spot labels from the gel images. Data is normalized / standardized by default and sorted as in the statistics table before being shown in the heat map.

Heat map for the example project.

Discovering patterns in expression profiles

Clustering of images is a good first step in assessing the quality of the quantitative data.
Employ 'hierarchical clustering' to show more structure in the data: Press the HCL button in the toolbar. Choose 'Gene Tree', 'Euclidian Distance' as metric and 'Complete Linkage'. To confirm press OK. The hierarchical clustering now groups expression profiles in accordance with their similarity.

Dialog box for HCL settings.

The tree represents similarities between expression profiles. Click into a subtree to select all protein spots of the same cluster. Since the statistics module is also synchronized with e.g. the Quantitation Table and the Dual View, you can switch there and review the selected spots as members of the selected cluster.

If you wish to identify the structure of your experiment choose 'Sample Tree' as tree selection. All replicates of the same sample should appear in the same cluster (subtree).

Hierarchical Clustering for Expression Profiles.

Finding profiles for defined Templates

With Template Matching, you can define a template for an expression profile and let Delta2D find spots whose expression profiles match the template.

Click the PTM button to open the dialog box. The lower part of the dialog contains a series of sliders. Move them roughly to different positions to define a sample expression profile that shall serve as a template and confirm it with OK.

Expression Profiles matching a Template

TMEV displays a new Heatmap containing the defined template and a list of expression profiles that match this template.

Finding differentially expressed proteins: Statistical Tests

Delta2D provides methods to control the proportion of false positives in the result set (False Discovery Rate - FDR). Overall, the False Discovery Rate approach allows one to strike a balance between the need to find statistically valid proteins of interest and the additional cost that is associated with following up on false positives.

In the t-Test options dialog, select "p-values based on permutations" and "Stepdown Westfall and Young methods". Choose bounds for the number of false positive spots in the result set using the "number of false positive genes should not exceed". Alternatively choose a bound for the proportion of false positive spots in the result set, using the other radio button and text box.

t-Test parameters.

There is a lot more to know about statistical analysis in Delta2D. Please read on in the manual.