Data analysis: A step by step manual.
Previous Top Next

For fragment analysis two major steps are necessary:
1. Peak detection (Where are DNA fragment bands?) and
2. Sizing of the detected peaks (What base size has each DNA fragment).

Please follow the steps below to perform a DNA fragment analysis using GelQuest software:

Table 1. Typical file formats generated by fingerprint analysis are:
1. Data import
Prior to data analysis you need to import your trace data or gel images.
2. Sample selection, viewing and scaling
Sample selection:
In GelQuest's main window select the samples with a right mouse click on the sample name or select all samples using the menu 'Selection' - 'Select all'.

Show only size standard lane:
To carry out the following steps it is easier if you turn off all color channels except the size standard channel: Use the channel selector to view/hide individual color channels:
Channel selector

In the following text we assume that a 5-color system is used and that the size standard is in the orange channel.

View curve data:
If not already done, click the icon "Display curve data" to see the curves. This view is better for peak detection than the pseudo-gel image:
Display curve data icon

Optimal scaling:
Double click on the scale bars (right to the traces):
The scale bars are shown to the right of the trace data.

This will open the scale window. In the scale window adjust the Y-scale values for the orange channel to a value that the standard peaks hit the roof of the graphics.
Click the 'Apply to all samples' icon.
Click 'OK' button to close the Scale window.
3. Start analysis
Start analyis:
Click the 'Analyze' icon:
Analysis icon: Start sample analysis.

The display will automatically change to 'Scale by data points' and to 'Display raw data'.

View entire run:
Click the 'zoom out' icon repeatedly until you see all standard peaks of the samples:
Zoom out icon. Click repeatedly to view entire run.
4. Peak detection
Set default values for peak detection:
If you are not used to the analysis of trace profiles it is a good idea to have some typical values to start with the analysis. Click the 'Set Default values' button to set the factory settings for the analysis parameters. This will also carry out a first peak detection:
Set default values for peak detection.

Now, look at the trace data in the main window. Please check whether the size standard peaks were detected.
Frequently, too many peaks are detected because the factory settings for the peak detection can also detect minor peaks. In this case modify the peak detection parameters (see below).

Setting first and last data point:
If you have some unusual peaks at the beginning of the run (e.g. primer peaks) you may wish to cut them off, i.e. to restrict the analysis to a region in the center of the run. For this reason you may set the first and last data point for the analysis algorithm. These data point values refer to the raw data of the traces!
Please make sure that the samples do not vary too much in their peak locations because setting the 'First Data Point' and 'Last Data Point' value will be applied to all selected samples. If the samples differ too much it is recommended to analyse them seperately.

Setting and modifying the Minimum Peak Height values:
If too many peaks were detected you may try to obtain less peaks by raising the value for the 'Min Peak Height'.

Tip: To find the right data value you may double click on the traces in the main window to show a helper window with details on the peaks. Move the mouse over a typical size standard peak. The peak detail window will show the 'Height' value of this peak.

Back to the 'Sample analysis' window raise the 'Min Peak Height' value to a value that only the real size standard peaks are detected: Either use the up and down arrows of the table cell control or type an appropriate value into the table cell. A normal value would be 50 or even higher for ABI machines. If you enter the value using the keyboard and not the table cell control press 'return key' to confirm it (or use the 'Detect Peaks' button). If you wish to adjust this value to all color channels then click the 'Apply to all channels' icon in the table cell control.

If you are an expert you can also change the following parameters (if not set a zero instead):

Setting the Min. Peak Width:
This causes the software to detect peaks that have a minimum width of 'Min. Peak Width' data points. This ignores all peaks that are slimmer.

Setting the Min. Peak Area:
This will only detect peaks with an area greater than the defined value.

Min. Height-to-Width-Ratio:
Use this parameter to discriminate 'real peaks' having a prominent height from 'false peaks'.

Please note:
There is no general definition for a 'real peak'. However we could define that a peak is a 'real peak' if its height is more than twice its width.
5. Size standard matching
In the lower panel of the sample analysis window select the 'Size Standard Matching Parameters'.

Choose the size standard and on the right-hand side select the color channel for the size standard (here: orange Dye 5).

If you are new to GelQuest or use different size standards as offered by the software, you may define a new size standard template using the button 'New Standard'.

At least, click the 'Match Standard' button.
If too many peaks were detected for the size standard lane of certain samples GelQuest (since version 1.3) will not analyse these samples because the size standard matching could be very time consuming and, in addition, erroneous. In this case GelQuest will show a warning message. To analyse these samples: Select only these samples and repeat the steps for peak detection and size standard matching. It can be a good idea to analyse each sample separately.
5. In the main window:
Evaluate the result of the peak detection and size standard matching algorithm: Click the 'S>' icon to display the trace data (as curves) scaled by

After an analysis, please check, whether the size standard peaks of the size standard channel (here the orange ones) are aligned between different samples!

If not: Repeat the peak detection and try different peak detection parameters (especially modify the first and last data point) for those samples that were not aligned properly.