Home / code
Name Modified Size InfoDownloads / Week
Parent folder
lonestar_test.m 2013-06-04 4.7 kB
0
lonestar_train.m 2013-06-04 13.9 kB
0
LICENSE 2013-06-03 18.1 kB
0
README 2013-06-03 1.5 kB
0
Totals: 4 Items   38.2 kB 0 
---------------------------------------------------------------------------
Step 1: Obtain MATLAB cvx toolbox.
---------------------------------------------------------------------------
        Download it from http://cvxr.com/cvx/download/ unzip
        or untar it in some directory say CVX_DIR. Run CVX_DIR/cvx_setup.m


---------------------------------------------------------------------------        
Step 2: Train lonestar algorithm
---------------------------------------------------------------------------
        Lonestar is a supervised binary classification and feature reduction 
        algorithm. It requires: 1) input data, a real matrix of size say m-by-n, 
        where each row represents features (genes) and columns represents 
        samples. 2) a binary class labels associated with each of n samples. These 
        class labels should be a n-dim vector with values of either 1 or 2.

	Depending on size of input this step may take long time to finish.
	For a typical case when input is 12K-by-200 with about 800 statistically significant features, it takes around 20 minutes on a standard desktop (say i5-2540M CPU)

        Please see detailed description under lonestar_train.m

---------------------------------------------------------------------------        
Step 2: Classify new samples with the classifier produced by step 2.
---------------------------------------------------------------------------
        Please see detailed description under lonestar_test.m
Source: README, updated 2013-06-03