BASE Test procedures Performance

Performance testing of BASE

1. Running the test pgoram

The data files needed by the tests are NOT included in the subversion repository. The main reason is that they are too large, and that we don't have permission to make them publicly available for download. To get the test file you need to be a core developer. Read the instructions on the DeveloperInformation page, Test data section on the Base 2 web site.

To run the test do the following:

1. Compile the core and the test programs: ant test.
2. Change to the build/test/ directory.
3. Run test program:

   ./test.sh performace <options> <cmd>
   

Test program options

• global options:
  • -u: Username to run the test as. The default value is "root".
  • -p: Password for the user. The default value is "root".
  • -x: If present, the tests are executed immediately. Otherwise they are added as jobs to the job queue.

• raw: import raw data
  • -i: Number of raw bioassays to create. The default value is 1.
  • -c: Name prefix for the created raw bioassays. A running number is appended to the name automatically. The default value is "Raw".

• root: create root bioassay sets
  • -s: Name prefix of raw bioassays that should be used. If no value is given, all raw bioassays in the experiment are used.
  • -c: Name prefix for the created bioassay sets. The iteration number and number of raw data sets is appended to the name automatically. The default value is "Root".
  • -i: Number of root bioassay sets to create in each iteration. The default value is 1.
  • -mi: Number of source raw bioassays to use in the first iteration. The default value is 1. The number of source raw bioassays are double in each iteration.
  • -mx: Number of source raw bioassays to use in the last iteration. The default value is 1.
  Note! With the -i, -mi and -mx parameters it is possible to specify that multiple root bioassay sets are created. Here is an example: -i 2 -mi 10 -mx 80. This creates 2 root bioassays with 10, 20, 40 and 80 source raw bioassays each.

• filter: create filtered bioassay sets
  • -s: Name prefix of bioassay sets that should be filtered. The default value is "Root".
  • -c: Name prefix for the created bioassay sets. The default value is "Filtered".

• lowess: create normalized bioassay sets using lowess
  • -s: Name prefix of bioassay sets that should be normalized. The default value is "Filtered".
  • -c: Name prefix for the created bioassay sets. The default value is "Lowess".

• export: export bioassay set data
  • -s: Name prefix of bioassay sets that should be exported. There is no default value.

• clean: delete all test data, except reporters

The test doesn't program doesn't require an empty database. Running the test program multiple times will simply add more data to the experiment.

Parameters to plug-ins are read from the performance.properties configuration file. It should be located in either /config/test or /config/local directory. If you want to change the parameters put the changed file in /config/local. Don't put this file into the subversion repository.

2. A typical test sequence

In this case we assume a fresh, empty database.

# Import raw data to 10 raw bioassays.
./test.sh performace -i 10 raw

# Create 2 root bioassay sets each with 1, 2, 4, and 8
# raw bioassays
./test.sh performance -i 2 -mi 1 -mx 8 root

# Filter all root bioassay sets
./test.sh performance -s Root -c Filtered filter

# Normalize all filtered bioassay sets
./test.sh performance -s Filtered -c Lowess lowess

# Export all root bioassay sets
./test.sh performance -s Root export

# Export all filtered bioassay sets
./test.sh performance -s Filtered export