Method 1 - Complete gel analysis

2D Gel Analysis - part 1: Obtaining spot coordinates

1. Open the ``reference'' image. This should be the best 2D gel available, showing as many spots as possible.
2. Clean up the 2D gel image to eliminate streaks, edges, molecular weight markers, labels, borders, dirt, etc. The algorithm can match spots correctly if as few as half the points correspond to actual spots, but matching will be more accurate if no noise is present.
3. Filter the image (using ``Force background to fixed value'' or ``Remove low frequencies'' if necessary to ensure that the background is constant.
4. If some of the spots are faint, make them darker by increasing the contrast or by drawing/painting on the spot. This will not affect the results, as the spot signals are ignored at this step. (Don't save the modified image, however. The original image will be needed for densitometry).
5. Using the ``draw'' function, draw white lines on the image to separate any spots that are overlapping.
6. If the `unknown' gel is rotated with respect to the reference gel, better results will be obtained if you rotate it manually before starting. This can be done in Image Registration or by clicking ``Image...Rotate''.
7. Click ``Grain/Pattern counting''
8. Perform Grain Counting on both images, using appropriate threshold and minimum size values so that all the spots are counted.
9. Click ``Save Grain Results'' to write the spot locations to disk.
10. Make a backup copy of the grain counting data. This will be used during densitometry.
11. Save a copy of the modified, marked image for future reference. (Don't overwrite the original image).
12. Repeat for the other ``unknown'' images.

2D Gel Analysis - part 2: Generating spot list

1. Click ``Image..Image registration''.
2. Read the data points for reference image. This is a list of all the spots that were found, obtained from Grain Counting in part 1.
3. Repeat for unknown image.
4. Save match table - if something goes wrong or you wish to reanalyze, you can just click on ``Read match table'' (optional).
5. Create a landmarks file by taking a subset of the most salient points from the reference and `unknown' data. (See below for format). These can either be a list of manually aligned spots, or an uncorrelated list of the most significant spots. The landmarks file should be a list of at least 6 of the most reproducible spots on each image, or at least 6 aligned spots.
6. Read the landmarks file. If the landmarks file is a list of manually obtained landmarks, proceed to step 8.
7. Correlate points - If the landmarks file is a list of the most significant spots, but the correspondence between the spots is unknown, they must be correlated before generating a vector map. Imal will attempt to calculate the correspondence using pattern-matching. The spots from the unknown image will be reordered so that the best statistical matching is shown. As the number of landmark spots increases, the difficulty in determining which unknown spot corresponds to which reference spot increases by n$^4$. Thus, best results are obtained if the total number of landmarks to be determined from each image is kept below 100.
8. Calculate vector map - uses landmark spots from the previous steps to create a mapping of the `unknown' image onto the reference image.
9. The calculated shifts will be shown on a new image, with changes in red indicating horizontal shifts and changes in blue indicating vertical shifts. Thus, the color code would be:
   • Green = large left shift
   • Red = large right shift
   • Yellow = large up shift
   • Blue = large down shift
   Mixed vertical and horizontal shifts are indicated by intermediate colors. The landmark points and their vectors are superimposed on the map.
10. Referring to the displayed vector map, click ``Edit matching table'' and edit the match table to remove any incorrect links (see figure below). The vector map should be a smooth gradient of colors. All the displayed vectors should be roughly parallel to the vectors nearby. Any points containing two or more vectors should be inspected and the incorrect link removed in the editor. Any vectors wildly different from their neighbors should be inspected and removed if necessary.

    
    
    

    Part of a vector map showing an incorrect link that should be removed. The vectors point from landmark points on the unknown image to landmark points on the reference image. The numbers are user-specified point labels; black labels are from the unknown image, and white labels are from the reference image. In this example, the data files were created by grain/spot counting, so the labels are all numbers. The link from 11 to 28 is incorrect because it has a markedly different size and angle than its neighbors. The ``28'' is also shown in red because its coordinates were found in the landmarks file but not the data file. The link from 5 to 16 may also be incorrect.
    

11. Recalculate vector map and repeat the previous step until an accurate and reasonable vector map is obtained.
12. Once the correct matching has been obtained, click ``Save'' in the matching table editor to save the matches. This file can also be reloaded later if desired.
13. Create unwarped spot list - Changes the x,y coordinates of the data points from the unknown image to match the reference image. (Note that this does not warp the image.)
14. In the Spot List Editor, click `Save' to save the corrected spot coordinates for part 3.
15. The spot list can be used to create a composite image displaying the spots lined up in a rectangular array (sec.6.5.1).

2D Gel Analysis - part 3: Measuring the signals in each spot

1. Reload the original, unmodified image.
2. Click on ``Spot densitometry''
3. Set pixel calibration, maximum signal, calibration factor, and background value or automatic background as desired (see Sec 8.11 for details).
4. Select data source:
   1. Spot list from image registration (kept in memory)
   2. Disk file in `spot list' format (see below for details)
5. Click `Accept'. imal will automatically perform densitometry on each spot and create a list.
6. Click ``Save'' on the editor box to save the results.

Alternatively, you could use Grain Counting or Auto Find Spots (Sec. 8.11.2) to quantitate the spots.

NOTES

• The data points should be loaded before creating the vector map from the landmark points in order to estimate the correct size to make the vector map. Otherwise, the vector map may be too small. If this happens, load the data points and click ``Calculate vector map'' again.
• The distance metric used in creating the vector map is appropriate for images of objects that have undergone elastic deformation or some equivalent shape change, such as 2D polyacrylamide gels being warped by temperature gradients. Some other types of images may require a different method for calculating the vector map. Contact author for assistance.
• Deleting links: If the numbers of points in the two data sets are unequal, some of the data points in one set will be closest to two or more points in the other set. Although most of these extra links are automatically pruned, imal is very conservative about pruning links. This means that occasionally, a data point will have 2 or more vectors pointing to it. It is usually obvious by inspection which one is incorrect. Deleting the entire line from the match table editor will remove the incorrect link and the incorrect data point. If you wish to keep the incorrect data point but remove only the link, change the x and y coordinates of its matching point to zeros. On the next recalculation, the point will still be plotted, but will be listed as an ``Unmatched point''.
• Currently only rectangular region spot densitometry is supported. It is recommended to set `Auto bkg/fuzzy k means' if there is an uneven background.
• The editor has a maximum capacity of 10,000 spots and 384K characters. These limits can be increased by changing RPOINTS and EDITSIZE, respectively, in xmtnimage.h and recompiling.