A common challenge for researchers is combining their clinical data
with other sets of data. In this step we will show how LabKey Server solves this common data integration problem. We will combine two different datasets into one and then build visualizations based on the combined result.
Examine the Two Source Datasets
- Click the Clinical and Assay Data tab.
- Open these two datasets (from the "Clinical Data" section) in different browser tabs so you can view and compare side by side.
- Physical Exam: This dataset captures the vital signs of the study participants: blood pressure, pulse and respiration rates, etc.
- Lab Results: This dataset captures the laboratory work done on the blood samples provided by the participants: lymphocyte levels, hemoglobin, etc.
Together, these two tables should give a comprehensive picture of the participants' hematological health and there may be relationships that can be detected in the combined data. For example, is there a relationship between the blood pressure data (in the Physical Exam set) and the lymphocyte levels (in the Lab Results set)? Or other relationships? To answer these questions we need to put all the data in one bucket somehow. How do we combine these two tables so that we can see all of the information in one grid?
Create a Combined Grid
Here we create a joined grid view combining all of the blood-related data:
- In the Physical Exam dataset, select (Grid Views) > Customize Grid.
- In the Available Fields pane, scroll down and click the (expand) icon next to DataSets. Note: "Datasets" is grayed out because it is not a "field" you can include in your grid.
- Scroll down and click (expand) next to the Lab Results dataset (also grayed out).
- Check all of these blood-related fields: CD4+, Lymphs, Hemoglobin, Viral Load.
- Scroll back up and remove checkmarks from: Clinician Signature/Date, Pregnancy, Form Language (to remove clutter).
- Click Save, select Named and name the grid "Hema/Cardio Data".
- Leave the box checked to Make this grid view available to all users.
- Click Save.
Create a Visualization
We now have an integrated (joined) grid view of the all of the participants' hematological data, which you can return to later by name under (Grid Views) > Hema/Cardio Data
. Values from the Lab Results dataset are added to the Physical Exam dataset, if values are available for the same participant and date combination. Now we can start making soundings into this combined data to see if there are any relationships to be discovered.
First let's create a scatter plot to see if there is a relationship between the lymphocyte levels and the blood pressure levels.
- If necessary, return to the Physical Exam dataset and select (Grid Views) > Hema/Cardio Data.
- Select (Charts/Reports) > Create Chart to open the chart wizard.
- Click Scatter on the left.
- Drag and drop Systolic Blood Pressure in the X Axis box. This column came from the Physical Exam dataset.
- Choose Lymphs (cells/mm3) as the Y Axis. This column came from the Lab Results dataset.
- Click Apply.
- The scatter plot is displayed. (A quick visual check suggests there is no relationship, at least not in this fictional sample data.)
- Save the plot with the name of your choice.
- Experiment with the chart tools to see if you can discover any relationship within the data.
- Click Edit to reopen the chart tools.
- Chart Type lets you add grouping or point shaping by other columns, such as cohort or demographic information.
- For instance, try dragging the "Participant Group: Treatment Group" column to the "Color" field.
- Click Apply to see the chart with the color coding. Notice that the lowest Lymph levels seem to belong to participants in the HIV+ group not receiving ARV treatment.
- Chart Layout provides more options for changing the chart title, size, coloring, grouping by density (in a heatmap), etc.
- You can Save overwriting the previous version, or Save As a new chart.