2.31 MB

Create the Consensus Spot Pattern

Delta2D's 100% Spot Matching is based on advanced image processing methods. It was introduced by DECODON in 2003. For a general presentation of this approach, please go to "100% Spot Matching".

100% Spot Matching produces complete expression profiles for every protein. Besides higher throughput, this leads to significantly improved statistical confidence, so you can, for example, identify more biomarker candidates from the same experiment. Other approaches to spot detection and matching lead to inconsistencies like missing values in expression profiles, and ambiguities in the profiles themselves.

Based on a completely warped project, the quantitative analysis in Delta2D consists of three steps:

1. Image fusion to produce an image that shows all spots in the experiment.
2. Spot detection on fused image to produce a 'consensus spot pattern'.
3. Transfer of Spots to the original images where they are quantified.

Generating a fusion image

The fusion image shows all spots of the experiment and serves as the basis for consistent spot detection and matching. Image fusion works by warping a set of images and then combining their intensities pixel by pixel into a new image. The result is an artificial gel image that has realistic spot shapes and combines essential characteristics from the original images.

There are many interesting applications of image fusion. For example, fusion images can be used to create a 'proteome map' i.e. a condensed visualization of all analysis results where you can collect your spot annotations. With average fusion, experimental variation can be compensated by combining several replicates into one average image. Large numbers of gels can be reduced to a single representative one.

Let us create a fusion image of our experiment: Switch to the fourth step in the Workflow window called 'Create the Consensus Spot Pattern' and click on the 'Fuse all images...' link to open the 'Image Fusion' dialog: Select 'control_01'as 'Master Gel Image' and 'Union' as Fusion Type, since the union fused image will include all the spots of the experiments. Leave all of the other settings. Click on 'Fuse' and a new image will be created and placed into a new group called 'Fused Images'.

Open the new fusion image by clicking on 'Open Fused Image using Union...' in the fourth Workflow step. A Dual View window opens showing the single view of the fused image only.

If an image is not connected to any of the other images or is only connected with the 'automatic' warp mode and you have not started the Automatic to find match vectors so far, you won't be allowed to fuse those images. If necessary, please check your warping strategy in the 'Warping Setup' window ('Window' -> 'Warping Setup') once again.

You can see that the fusion image looks like a real gel image. The natural spot patterns result from combining the images pixel-by-pixel using a weighted average function. Most importantly, the 'Union' fusion image contains all spots of the project's gel images, even those that occur only on one or two of the images.

For DIGE projects you can exclude the internal standard images (typically the Cy2 labelled images) from the fusion since they do not carry information which is not included in the sample images.

Detecting spots on the fusion image

Keep the Dual View open and click on the link 'Detect Spots on Fused Image...' in the 'Create the Consensus Spot Pattern' in the Workflow (alternatively choose 'Spots' -> 'Detect Spots on Fused Image' in the Dual View). A new dialog opens where you have access to the spot detection parameters.

Just accept the proposed parameter values since they are quite reasonable in most cases. Confirm this dialog by pressing OK and let Delta2D detect the spots.

The default parameters for the spot detection are derived from the actual images and result in a very sensitive spot detection. This is because it is much quicker to filter out false positive rather than adding new spots one-by-one.

Editing spots on the fusion image

You can correct the results of Delta2D's automatic spot detection by setting "markers". By using markers you can control where a new spot should be detected. Delta2D will then compute the new boundary accordingly. There are three basic operations for spot editing: adding a new spot, splitting a spot, and joining two or more spots. In any case Delta2D will compute spot boundaries automatically according to your input. This approach to spot editing maximizes reproducibility while allowing you to refine the detected spot pattern.

In order to edit spots, you have to select the 'Spot Editing Tool' first. Click on the third button in the vertical tool panel at the left of the Dual View window.

Adding a New Spot on the fusion image

To create a new spot simply click on the center of the undetected spot to let Delta2D automatically find a spot at this position. A new spot marker (shown as a "+" symbol) is created and Delta2D will compute the spot's boundary.

Image region before and after adding a spot manually

Joining Spots

To join two or more spots, connect the spot centers with a line-marker: just click and drag with the mouse. Delta2D uses this marker to compute a new spot boundary that covers the spots that have been connected by the line-marker.

Examples for joined spots.

Splitting a Spot

In some cases Delta2D detects one spot where you would rather like to have two or more. Click once into the center of the spot that was not detected separately and a spot-marker will appear. Starting from this marker Delta2D will now compute new spot boundaries. Depending on the position of the marker the spot boundaries will vary.

Set of spots, detected as one and split with markers

Moving and Deleting Spot Markers

Spot markers can be moved by dragging with the left mouse button. To delete a spot marker you can right-click on it. Moving or deleting markers changes spot boundaries and it may take some time to compute the new boundaries.

Removing Spots

To remove a spot, activate the 'Spot Selection Tool', right click on the respective spot and select 'Cancel Spot' from the context menu. To remove a complete region of false positive spots, for instance at the gel image borders, you can select groups of spots by dragging a rectangle around them and cancel them together. By selecting 'Spots' > 'Show Cancelled Spots' you can display the cancelled spots with dotted lines.

Filtering Noise Spots

While Delta2D usually is excellent in finding spots, some image properties such as strong image noise may cause the detection of 'false positive' or 'noise' spots. These segments do not have the typical shape when viewed in 3D, and they are generally quite faint.
You can use the 'Sensitivity' parameter in the Spot detection parameter dialog to suppress the detection of background spots. However, Delta2D also computes a 'spot quality' value for every detected spot. This value shows how closely a spot represents the 'ideal' 3D Gaussian bell shape. You can filter spots in accordance to this value like this:

1. Open the quantitation table by choosing Window -> Quantitation Table.
2. In the table, click on the tab for the fused image.
3. Locate the spot quality column, and click on the 'filter' button in the column header.
4. In the filter dialog, check the 'Filter active' check box, then use the sliders to adjust the filter settings or insert values as filter borders (a value of 0.02 works fine in may cases).
5. Press the Apply button to check the effect of your filter immediately, or click OK when you are done.
6. Use the Dual View to check the spots that have been filtered.

We recommend setting the filter in such a way that you see only those spots that you would like to remove. Then choose 'Edit->Select all' and 'Cancel -> Cancel Selected Spots'.
Finally, to remove the filter on the spot quality column, click on the filter button in the column header and uncheck the 'Filter active' checkbox. Now you have cancelled all spots with insufficient spot quality.

In summary we have detected spots on the fused image and since the union fusion in-cludes all spots occurring on any of the gel images, we have a 'consensus spot pattern' for the whole project located on the fused image.

Transferring Spots

The next step towards quantitative analysis is to transfer the consensus spot pattern from the fusion image to the other gel images where the spots will be quantitated automatically.

In the step 4 'Create the Consensus Spot Pattern' in the 'Workflow' window click on the fused image 'Transfer Spots...'. The 'Transfer' dialog opens. The default settings will be that spots are transferred from the fused image to all the other images of the project. Confirm these settings with OK.

If an image is not connected to any of the other images or is only connected with the 'automatic' warp mode and you have not started the Automatic to find match vectors so far, you won't be allowed to transfer spots to that image. If necessary, please check your warping strategy in the 'Warping Setup' window ('Window' -> 'Warping Setup') once again.

A progress bar indicates how spots are transferred to one image after another and quantified. Since we started with the detection result from a union-fused image we get a complete expression profile for every spot over the whole set of gel images.

Spot boundaries after spot transfer. Choose 'Window -> Gel Image Regions' to open this view. Please note that the essential spot patterns are the same on all gels, making 100% spot matching possible.

You can choose either to transfer the boundaries as they appear on the fusion image or to adapt the size of the boundaries to the actual spots on the target images. Find this option in the Preferences tab 'Spots'.

On those images where a certain spot does not contain a signal, Delta2D will nevertheless quantitate the images in accordance to the transferred spot boundary. This will result in a spot with near-zero quantity. You have transferred the information where you expect the spot in the single gel images.

In order to see the quantitative expression profiles choose the menu entry 'Windows' -> 'Quantitation Table'. This will open the Quantitation Table.

Before looking at the table in more detail let us have a quick look at some more visual means for the analysis of expression profiles.