ROC combine
This software computes the ROC curve AUC off a combination of all parameters given in a CSV Table. When the AUC is above a threshold, it draws the corresponding ROC curve. Caution : do not process large tables with this software ! With 10 parameters the number of combinations computed is about 5000.
Acknowledgment : I thank very much the mathematician Sebastien Dejean who suggested me to combine the parameters using the logistic regression.
Manual :
1- if not already installed, install R programming language and GNU parallel.
2- install package pROC : install.packages("pROC")
install package snowfall : install.packages("snowfall")
3- unzip the software
4- copy your files in the “data” directory. Very important : avoid spaces or special characters in files names. The files must be formated as follows :
- txt or csv files with TAB separator
- first line : column names
- first column : row names
- second column : label this column 'group' and the values should be 0 for one group and 1 for the other one.
- next columns : values with dot as decimal separator.
For clinical studies: patients are in lines and clinical criteria in columns
3- execute the software by the command : perl parallel_ROC-0.1.pl
4- plots are in the "svg" directory and tables with AUC and optimal cut are in “results” directory.