MBBC tutorial
Step 0
Please download and extract MBBC20zip.exe.
Step 1
Step 1-1: Run MBBC.exe
Step 1-2: Please check the installation of R and Ox.
If they are not installed properly, the "software has not been found" will be displayed.
If the "software has not been found" appears even after proper installations,
please set up the path of R.exe and oxl.exe manually.
Step 1-3: Default output directory is assigned. You may change it to wherever you want.
Step 1-4: You may use the 'browse button' on the right side of line edit window to load a data file,
or may type the data file name with the whole path
(e.g., C:\Documents and Settings\circinus\My Documents\wound.txt).
For your convenience, the 'Example' menu provides the default set up of
configuration information and data file. It would be helpful for you.
Step 1-5: After the data file name is entered, the number of time points,
genes, and replications must be specified.
'Number of Genes' is automatically specified with the row size of data file.
'Number of Time Points' * 'Number of Replications' is the column size of data file.
Among three numbers, user specifies only the 'Number of Time Points'.
If you change the 'Number of Time Points', then the 'Number of Replications'
will be automatically adjusted.
The minimum value of 'Number of Time Points' is 2.
Step 1-6: The user has three profile-specific registration options: no registering, centering,
and standardization.
Step 1-7: Now, we completed preparation for loading the original data set.
Please click the 'Load Data' button.
Then, all registered gene profiles will be drawn as follows.
Step 2
To speed up the algorithm while still obtaining genetically meaningful clusters, we suggest reducing the number of genes to be
clustered ('Number of Remaining Genes after Filtering') using the polynomial regression models over time. The range of 'Number of Remaining Genes after filtering' must be between 1 and 'Number of Genes'.
If 'Number of Remaining Genes after Filtering' is less than 'Number of Genes', 'Remaining Gene Profiles after Filtering' will be drawn on the right panel.
Now, data set is ready for cluster analysis.
Step 3
Step 3-1:
The 'Number of Iterations' must be set between 10 and 900000000.
The higher number of iterationss, the higher chance of finding the optimal partition with the given finite number of iterations.
Before running the search algorithm with a large number of iterations, we recommend a test run with a small number of iterations
to examine the convergence and expected simulation time.
Step 3-2:
'Tuning Parameters' could be set by choosing either 'Log(m)' or 'Mean of Crowley's prior'.
See the reference paper for the details about 'Tuning Parameters'.
Step 3-3: 'Initial Partition' can be specified among user-defined partition in a file ('Open File'),
'Uniform random partition', and k-mean clusters. For details, see the reference documents.
Step 3-4: The smoothness of cluster-specific profile mean function is controlled by lanbda1 and lambda2. They can be either estimated or specified by the user.
Step 3-5: Now, it is ready to find the optimal partition.
Click the 'Search for the Optimal Partition' button to execute the search algorithm.
Step 3-6: The expected elapsed time is be calculated.
Step 3-7: After completion, clustered profile results and simulation history results
will be displayed on the right panel.
PDF report files containing all the configurations and profile results
as well as image files will also be created.
Step 3-8: By clicking the 'view report' icon, PDF file could be opened by
Acrobat Reader(c) or other PDF viewers.
