xQTL workbench - Setup a new runnable analysis
Create a parameter set
Similar to creating a DataSet definition, you can define a ParameterSet. You can configure this by clicking the Configure analysis tab in the main menu, then go to Parameters. Again, the structure is as follows:
- Analysis: the complete definition an an analysis, including data, parameters and function call.
- which has one ParameterSet: the wrapper for a set of parameters which can be used in this analysis. (can be shared across analyses) For example: 'Rqtl_params'.
- which has multiple ParameterNames: the 'types' of parameters used. For example: 'method' or 'stepsize'.
- which has multiple ParameterValues: the possible values which can be filled in for this parameter. For example: 'yes', 'no', 'Scanone', ScanMQM', '2', '5', '10'.
Add the analysis
- In the main menu, click Configure analysis and then Analyses.
- Click on Add new record. Give the new analysis a name and a description.
- Select which DataSet and ParameterSet you wish to use. Each can be reused individually.
- Set the TargetFunctionName. See below.
- Click Add to save the new analysis.
- An analysis must have a TargetFunctionName. This name is used to 'pick out' the R function to be executed when running this analysis.
- The convention is that any of the defined scripts has a function named 'run_' + TargetFunctionName.
- For example: You specify TargetFunctionName with value 'MYFUNCTION'. When the R api is loaded on execution, the framework will look for a function named 'run_MYFUNCTION' in the stack of sourced scripts. It does not matter if the scripts are part of the framework, or user defined.
- In the example data, the TargetFunctionName for Rqtl_analysis is 'QTL'. One of the sourced scripts, 'QTLjob.R', has a function named 'run_QTL'. So one is selected as the target to run. Notice the name of the script, 'QTLjob.R', seems a convention as well. While not strictly the case, naming your script thusly may prevent confusion or error.