TracNav(xQTL)?

xQTL workbench - Tutorial for biologists: Browse and interpret results

After clicking the link in the Output column in the previous part of the tutorial, you are taken to the results of the analysis. You'll nice that browsing the results takes place in the same data matrix viewer as mentioned before in the tutorial at xQTLTutorialBiologistInspectingData and in the general manual at xQTLBiologistBrowse. Here we will discuss a few things specific for looking at QTL profile results.

The first thing to notice is that the columns are typed as 'DerivedTrait'. The QTL analysis tool is not aware what specific type of trait is processed, so it uploads the results as 'DerivedTrait'. This type is normally used for a trait constructed during analysis, for example how one could construct a DerivedTrait called 'BMI' using Measurements 'Weight' and 'Length'. In any case, the source data were probes, so let's change this to Probe.

Click on Show additional fields. Find the dropdown box labeled ColumnType, and select Probe where it said DerivedTrait. Press save the changes. (the floppydisk icon)

Notice that the probe name ("A_06_P4350", "A_06_P7031", etc) turn from black to blue. This means the data matrix is now matched to the annotations. You can hover over these names to see more information, or click on them to immediatly browse to that record in the database in the Annotations menu.

Let's find the probes with very significant QTLs. You can filter the matrix to do this. In the matrix viewer, click on Action, then tick RC Filter on two dimensions. (in the Special column) Now fill in the filter: Select all probes with at least 1 marker(s) having a value greater than 10. Click Apply to all.

The visible area is now adjusted to display the probes for which this criterion is met, while showing all of the markers. So in this case you end up with 8 probes, times 282 markers.

Get a feel for the QTLs by plotting the filtered data. Click Action again, and tick Graph / heatmap with R. In the dropdown for Heatmap plot, select Just cols. This will cause R to cluster the data on the probes, so you can see which ones have similar QTL profiles. The marker order remains untouched. Set the resolution to 600*800, and press Plot visible. This means you plot what is currenly in the visible area, which is our filtered data.

Open the plot by click on the thumbnail picture that appears. Which probes share similar QTL profiles?

You may want to download a copy of these results for future reference. Click the Download button. In the column that says Visible values, click on Excel file. If you try CSV format or R matrix object, use the Back button of your browser to return to the matrix viewer.

Now, try using the two-dimensional filter to zoom further into the data. Select only those markers with at least 1 probe with a LOD score of > 5. Remember to click on Apply to visible. Try to create the same type of heatmap plot as before. Is this plot more useful?

Let's take a different viewpoint on the data. Press the Reset button. This will restore the matrix to its original, unfiltered state. We will filter the markers on their chromosome annotation. Click Action, under Filter on attributes: tick markers. Now set the filter as follows: Chromosome_name, equals exactly, chr15. Click Apply to all.

All of the probes have been selected, but only 31 markers: the ones on chromosome 15. You could plot this filtered set to see what's going on at this chromosome. If you now apply the same two-dimensional Select all probes with at least 1 marker(s) having a value greater than 10 filter as before, you are left with 3 very interesting probes on this chromosome.

After a lot of filtering and browsing you may want to store a data matrix view permanently. Click Download, Visible values, CSV format. Use your operating systems regular Select all (CTRL+A for Windows and Linux, CMD+A for Mac) and Copy (CTRL+C) functions to store the data to clipboard. Go back to the matrix view. Click the Add new record button. ("sheet of paper with green plus" icon) Fill in a name (e.g. MyData), and set the ValueType to Decimal. Leave the rest as it is.

If you are still looking at the 'source' matrix, remove the filter by clicking the red X. Browse to the MyData matrix via the controls underneath Navigate data matrices, or go to the List view and select it from there. (blue-yellow icon next to Add new record) Clear the box with Alternatively, use this textarea to input your data. over it, and use paste (CTRL+V) to enter your data. Now press the Upload button located on the immediate right-hand side of this box.

Your data selection is now stored as a new matrix. Notice that the annotions are wrong, which you can fix by now. In addition you probably will want to explain something about this piece of data, using the Description field of the new matrix entry. (use Show additional fields)

Previous: Running a basic QTL analysis