Supporting page of "ROS-DET: Robust detector of switching mechanisms in gene expression"


Binary files of ROS-DET are available.
This page shows a brief summary for running the binary file.
A readme file is available here.

Download

Each of the following compressed files includes a binary file ("rosdet"), sample files (exp.txt, label.txt, out_sample.txt) and a readme file.

Platform File
Linux (x86_64) rosdet_linux.tar.gz (test version)
Apple Mac (i386, PPC) rosdet_mac.dmg (test version)

The result of the sample input ("exp.txt", "label.txt") was output in "out_sample.txt".
The input file ("exp.txt") include synthetic data of expression values of 8 genes for 30 individuals.


How to run

Example
Please type the following command at a terminal window of Linux and Mac.
     >> ./rosdet -e exp.txt -l label.txt -o out.txt 
     >> significant level for gene pairs to be selected: (enter a significant level, for example 0.001)
     >> number of gene pairs to be output: (enter a number of gene pairs, for example 100)

The result of the input ("exp.txt", "label.txt") is output in "out.txt".

Options
'-e' specifies a data file for expression values of genes "exp.txt".
'-l' specifies a data file for labels "label.txt".
'-o' specifies an output file "out.txt".


Data format

Each of input files contains numerical data which are written as a matrix formulation:

data type
expression:
label:
 size (rows × columns)
#genes × n
1 × n
delimiters
white space / tab
white space / tab
 sample
"exp.txt"
"label.txt"

Where n is the number of the individuals.
Each label should be 0 or 1.
Both white space and tab are available as delimiters of data.
However, continuing spaces and continuing tabs are NOT ALLOWED.

Note: ROS-DET allows #genes up to ~12,500.
(our current confirmation using Intel XEON(R) CPU 3.33 GHz, 32 GB RAM)

Results

The result will be printed in "out.txt":
--------------------------------------------------------------------
--------------------------------------------------------------------
Results of ROS-DET:
-------------------------------------------------
significant level = 0.001786
number of gene pairs to be output = 100 (maximum)
-------------------------------------------------
rank score(WCOR) log10(p-value(ECOR)) gene X gene Y r1 r2 weight c
1 1.18 -10.83 7 8 0.86 -0.90 0.67
2 0.14 -9.07 5 6 0.88 -0.80 0.08
3 0.14 -6.61 3 4 0.95 -0.23 0.12
--------------------------------------------------------------------
-------------------------------------------------------------------- 

1st column: rank of gene pair
2nd column: score of WCOR (=c|r1-r2|) 
3rd column: log10(p-value) of ECOR
4th column: index of gene1 (=1, 2, ..., #Genes-1)
5th column: index of gene2 (=2, 3, ..., #Genes)
6th column: biweight midcorrelation for class 1(label=1)
7th column: biweight midcorrelation for class 2(label=0)
8th column: our weight c in WCOR