BRMatlab (v. 0.1)

Instructions

Launch Matlab

Donwload the file ProjBRMatlab.zip

Open the zip file in a directory of your choice

• List of file in the directory after unzip
• BRMatlab_Elab.m
• BRMatlab_Prepare.m
• Directory libPLS_1.98

In Matlab open the directory with the scripts using the window Current Folder and then add the folder and the subfolder to Matlab path using the mouse right bottom. The addition of the folder and subfolder is mandatory because Matlab need to know where the program is located.

In addition, it is also mandatory to add the folder where the data that will be submitted to BR Analysis is located.

Type in the Command Window

Then choose the descriptor matrix and the response file in the GUI

• In the two files the  data are separated by ";"
• Other separators were not tested
• There are no restrictions about the working directory

BRMatlab_Prepare writes two files in the working directory:

• BRMatlab_DataMatrix.csv
• It will be read by BRMatlab_Elaborate.m
• TrainingTest.txt
• The file contains the name of the object and a numerical flag
• In the file the flag is equal to 1
• Change the flag value to 0 for compounds in the test set

BRMatlab_Prepare sets the environment thus it is mandatory to launch the script before proceeding with the elaborate step. When the environment is set it is possible change parameters in the elaboration step.

When the test set is defined, type in the Command Window

The script runs PLS and performs  BR analysis.

The cross-validation is performed using LOO method, but it is possible to change to K-fold method modifying the script in the cross validation section. Some options are included and commented with % symbol. The option can be activated removing % and adding a % symbol before instruction K = lenght(YTr).

The results are written  in a file named BRMatlab_Results.txt