Output

Through the Run panel it is possible to operate on the power study process (Start, Pause, Resume, Abort buttons) and to visualize the results in real time.

The LOD panel shows the Mean LOD-score at each position of the map. In a comparative analysis both sample results are graphed in this panel.

Above the graph it will be shown the 95% empirical confidence interval (CI) for the QTL localization (in cM) as observed during the simulation process. It is evaluated basing on the location of the highest peak in each simulation.

The H2 QTL panel graphs the LOD-score versus the observed QTL heritability.

This information allows predicting what could be the estimated QTL effect, independently from the generating value, when significant evidence of linkage is observed (for example see Goring et al 2001).

The LOD Freqs panel shows the histograms of the observed power at each map position, by counting the number of times the test statistic falls above the user specified thresholds.

Above the graph is reported the power observed in the whole region. Since chance variations in location estimates are expected (as in the real data), sometimes the test statistics can be higher at a map position far from the QTL location (see also the LOD panel for the CI). The power observed in the whole region better represents the real power for the study, and it is likely to be higher than the power observed at the closest markers.

The Log panel contains the details about each simulation in ASCII format and can be imported in a standard statistical package for further examinations.

The H2T panel shows the distribution of the observed trait heritability within the simulations.

Export button

At the end of the simulations, all the graphs and the Log data can be exported by pressing the Export button.

In each graph panel it is possible to choosing between an automatic scale and a custom scale for the visualized graph. When exported the pictures are the exact copy of what visualized in each panel and will be saved in JPG format.

Notes: computational charge

Even if particular attentions have been applied to the hope of minimize computational time and memory demand, depending on the sample size and the number of markers the simulation study can be relatively slow. During the simulations a progress bar estimates the time needed to end all the simulations. This time can fluctuate if there are other active processes in the computer (in a computer dedicated to the simulation study, after the first 10% of the total number of simulations the remaining time is quite reliable).

Heavy simulation studies can be performed overnight or in a dedicated computer. If a Windows PC needs to be used for other purposes while the simulations are running, it can be convenient to lower the PowQ process priority. In the latest Windows versions it could be done by pressing CTR-ALT-CANC, selecting Task Manager, Processes, right-clicking on the PowQ process (which should be named javaw.exe) and selecting Set Priority -> BelowNormal. Anyway, by lowering the priority the simulations will take a longer time to be completed.

Memory low: PowQ could runs out of memory. This is because of the way that Java runs on a computer - what is actually run is a program called a virtual machine (the JVM) which executes the java instructions. The JVM has limits on the memory that can be allocated to the java program - and you might need to increase them if you are working with particularly large genealogy or with a large number of simulations. In order to increase the amount of memory for PowQ, the program should be run from the command line by writing for example:

java –jar -Xms64m -Xmx256m PowQ.jar

This sets the initial and maximum memory size to 64mb and 256mb. The m suffix can be changed with g to represent gigabyte.