Castle
Cluster Prism®
DraculaTM


Table Of Contents

  1. Overview
  2. Public Files
  3. Uploading Cell Files
  4. Model Generation
  5. Applying Model To Array Set
  6. Single Array Artifacts

Overview

Dracula is intended to estimate gene expression from Affymetrix GeneChip microarrays. The inputs are sets of Affymetrix Cell (.cel) files. Outputs are tables of genes expression esimates based upon the GeneChip® probe measurements in the Cell files. Additionally Dracula may be used to analyze artifacts on individual arrays. The steps involved in estimating gene expression are as follows.

  1. Upload the files to be analyzed. These files must all represent the same array type.
  2. Generate a model representing the properties of the array type.
  3. Apply the model to a set of arrays.
  4. Gene expression estimates are downloaded and/or stored in the Dracula database.
Users have their own space where their uploaded files and generated models and analyses are stored. Cluster Prism also provides models for some array types. Cluster Prism generated models are available to all users while user generated models are only available to the user generating the model.

The addition of tools, to do analysis on gene expression estimates, is planned but is not yet implemented.

NB: While the programs are running, the browser may say something like "Connecting". This usually means that the program is running. Programs may take a few minutes to run. You will need to wait for confirmation that the program has completed. Otherwise you will lose whatever the program was generating.

Public Files

If you do not have any Affymetrix cell files, or do not wish to upload your own, you may use get some of the public samples by going to the File Apps. menu and selecting Get Public Files. Select the array set, to be copied into your account, from the pull-down menu and click . The array set is now available for you to use. Addition data sets are also available at the Gene Expression Omnibus (GEO) site. Files cannot be downloaded directly from GEO to this site but the user can download files from GEO and upload them to this site. If an "Error Opening File" message appears during the model generation, this usually means that the library file is not available for that chip type. The user can upload the library file for that chip type by clicking the option in the menu of Dracula. For security reasons, the only Chip Description (CDF) files being accepted are those in purely ASCII format. We are working on allowing other CDF files to be uploaded. In the meantime, if you would like other specific CDF files to become available, please contact me at peterdlauren@yahoo.com.

Uploading Cell Files

From the File Apps. menu, select Upload Cell Files. This results in the following form.

Choose Name for Uploaded File Set:

This initiates the uploading of a cell file set. Uploaded files must be Affymetrix CEL files (ASCII or binary). All files in a given set must be of the same array/chip type. The user assigns a unique identifier to the file set. This is done by typing the name, to assign to the file set, in the yellow box. This identifier must be unique among the user's file set names. The user may subsequently access the set by selecting the identifier from the list of the user's set IDs. Once the unique set identifier name has been entered, click the

button. A dialog that includes
Click here to upload files.
should appear.

When you click on that text a file selection browser window should appear. You may choose up to 20 files for upload at once. If you require a larger set, you may subsequently upload more files to the same set. To select a contiguous set of files, click on the first filename and then hold down the Shift key while choosing the last filename. To choose a non-contiguous set, hold down the Control key while selecting the names of each of the required files. No single file may be more than 100 MB in size. In order to load more files, please follow the following steps.
  1. Upload a set of files with a total size no greater than about 300-400 MB.
  2. Refresh the browser.
  3. Upload the next set of files with a total size no greater than about 300-400 MB.
  4. Continue this process until all of the required files have been uploaded.
  5. When all of the required files have been uploaded, click the button.
Uploading files may take a few minutes and the time is proportional to the number and, to a lesser extent, the size of the uploaded files. If all of the files are in the correct format then an alert box saying "All files successfully uploaded" will appear. Otherwise an alert box will appear, identifying the files that are not in the correct format. In either case, clicking the OK button of the alert box will return the user to the Dracula menu.

All of the uploaded files in the correct are then saved under the selected ID. If no valid files have been uploaded, the ID will be discarded. Once the files have been uploaded, accepted and saved they can be subsequently accessed for model generation and gene expression analysis.

Removing uploaded sample sets

Uploaded sample sets can be removed, from the user's account, by going to the File Apps. menu and selecting Remove Cell File Set. Select the array set, to be removed, from the pull-down menu and click . The name of the selected set is shown in the background and an alert box asks if you really want to remove the set. If you do want to remove the set click OK. If you do not want to remove the set click Cancel.

Model Generation

From the Model Apps. menu, select Construct Model. Select the array set, to be used to generate the model, from the pull-down menu . Note: The set must contain at least two arrays. Otherwise the model generated will be garbage. Click . The program then starts running and the run time is proportional to the number of arrays in the sample set. The program runs for typically 10-12 minutes with a sample set of 42 arrays. When the program is finished running, an alert box will come up asking you to click it to return to the Dracula program. The model has now been generated for that chip type.

Applying Model To Array Set

From the Model Apps. menu, select Apply Model To Array List. Select the array set, to analyze, from the pull-down menu . Click . The program then starts running and the run time is proportional to the number of arrays in the sample set. The program runs for typically about 7 minutes with a sample set of 42 arrays. When the program is finished running, an alert box will come up asking you to click it to return to the Dracula program. An analysis file has now been generated and may be downloaded. It is a tab delimited ASCII table where the columns represent the individual arrays and the rows the individual genes.

Download Set Analysis File

From the Model Apps. menu, select Download Set Analysis File. Select the array set, to analyze, from the pull-down menu . Click . A dialog box will open, giving you the option of either opening the file or seving it to disk. The file that is downloaded is a tab delimited ASCII table where the columns represent the individual arrays and the rows the individual genes.

Single Array Artifacts

From the Model Apps. menu, select Apply Model To Single Array. Select the array set, to analyze, from the pull-down menu . Click . You may now choose whether to view the original image from the Affymetrix cell file or the background image which shows the variation, of the probe array image, that is independent of the gene expression that is to be measured. Make this selection by clicking on the appropriate radio button from the following.

    Show: Original Image   Background Image    

NB: While the model does not have to have been generated to display the original image, it must have been generated to display the background image. Attempting to generate a background image prior to generating the model will result in an Error reading file message. From the pull-down menu , select the array that you wish to view. Once you have selected the array, click the button. If the requested image is already available then it will display right away. Otherwise there will be a delay of about 40 seconds (for the original image) or 45 seconds (for the background image) while it is being generated. The selected image will be displayed along with a color bar reflecting the look up table (LUT) in use. The numbers on the right are the background values at the corresponding ends of the LUT. The difference between these numbers is the background range. The background variation will always look dramatic on the image but its significance is reflected by the background range and how it compares with the expression estimates. Once you have finished viewing the image you can click

RETURN TO SET MENU

to return to the menu to select another set and array or return to the main Dracula menu. Alternatively, you may use the display controls, to the right of the image, to change the enhancement features (look up table,brightness, gamma and nonlinear transformations) of the image display. The image will always be regenerated when the user lifts the mouse from changing a display setting. Since the changes in brightness and contrast are occurring over the Internet, they tend to be slower than if everything is happening on a local machine. The user may select a different color look up table (LUT) and/or a (different) nonlinear transformation of the gray levels (logarithmic/exponential increase/decrease). Not all nonlinear gray-level transformations are supported with some LUTs. The check boxes associated with unsupported transformations are disabled as the LUT is selected. Some nonlinear transformations are incompatible with each other. For example you cannot have both a logarithmic and exponential increase. As a transformation is selected, its incompatible transformation is automatically deselected.

When you log out of a session, the settings revert to those that were last saved (or the default registration settings if the user has never saved any new settings). If you would like to revert to the last saved display settings without logging out, click the button.

Region of Interest (ROI) Tool

Perhaps the most important reason to view an image of the original array, particularly of the background, is to identify artifacts on the array. The ROI tool provides a way for the user to mask out regions of the array to be excluded from the analysis. [NB. While the ROIs can be defined, having them excluded from the analysis still needs to be implemented.]

To define the ROI, click on the button. This toggles in and out of ROI mode. While in ROI mode, you may define regions of interest and change the color LUT. You cannot adjust the brightness or contrast of the image while in ROI mode. To change the brightness or contrast, you must toggle out of ROI mode.

While in ROI mode, you will see the following.


Rectangle Ellipse Freehand

The rectangle radio button will usually be checked since it is the default. If this is the case and you click on the button a rectangle will appear around the middle of the image. The rectangle can be dragged by clicking the left mouse button in the interior of the rectangle and holding the mouse button down while you drag the rectangle to the desired location. The rectangle may be resized by clicking on, and dragging, the circles in the upper right and lower left corners. If you would prefer an ellipse, select the Ellipse radio button before clicking the button. The ellipse is moved the same way as the rectangle and resized, and reshaped, with the circles at the top, bottom, left and right. If you would like to draw freehand, click the Freehand radio button and click the button. To draw, click the left mouse button where you would like to start and draw the desired shape. On lifting the mouse, the freehand ROI will automatically close by connecting the end of the trace to the beginning. You are then given the option of saving the trace by clicking OK or discarding it by clicking Cancel.

If you change the color look up table (LUT) and click on one of the radio buttons, the color of the trace will change to a color that is more suitable to the new LUT.

If you would like to delete an ROI that you have already saved, hold down the control key and click on the ROI to remove. If you confirm its removal, it will be deleted.

In order to save the entire set of ROIs, you have defined for this image, for later retrieval click on the button. Currently, only one set of ROIs can be saved for each array image. If a set has already been saved for this image, you will need to confirm whether you wish to overwrite it. Also, if a set of ROIs has been saved for this image, the button will appear when you toggle into ROI mode. If you click on this button, it will restore the previously saved set of ROIs.