ChIP-seq Peak Intensity - Expectation Maximization (CPI-EM) algorithm for detecting cooperatively bound transcription factor pairs. Paper reference to be updated soon.
cpiEm requires scipy, numpy, pandas and argparse modules to be installed. cpiEm runs with both python 2.7+ and python 3+.
The syntax for running cpiEm.py --
./cpiEm.py -h
usage: cpiEm.py [-h] --input-file <input file> [--mixture <dist>]
[--output-file <output file>] [--num-iterations <NUM>]
optional arguments:
-h, --help show this help message and exit
-i <input file>, --input-file <input file>
Name of input file. Input should be a tab-separated
file with two columns consisting of primary TF peak
intensities (first column) and partner TF peak
intensities (second column). The partner TF is assumed
to be TF that cooperatively aids in binding the
primary TF to DNA.
-m <dist>, --mixture <dist>
Distribution to be used to model peak intensities. Can
be lognormal, gaussian or gamma. Default : lognormal
-o <output file>, --output-file <output file>
File to which output will be written. Default : <input
file>.cpi-em
-n <NUM>, --num-iterations <NUM>
Maximum number of EM iterations to execute. Default :
10000
##Sample run
This is a sample run of the cpi-em algorithm, using the test-input file in the repository --
./cpiEm.py -i test-input -m gamma -n 1000
Output :
Fitting a gamma mixture to data
Maximum number of EM iterations to run : 1000
EM converged in 260 iterations. Q_2 = -2275.7959452
Writing output to test-input.cpi-em
The first five lines of test-input are :
3.59342 20.07449
9.98265 6.6389
7.50183 2.82852
14.07073 7.64249
3.34819 3.8665
The first five lines of the output file, test-input.cpi-em, are :
3.59342 20.07449 0.767812671493
9.98265 6.6389 0.765781602103
7.50183 2.82852 0.72318188783
14.07073 7.64249 0.743167549103
3.34819 3.8665 0.667208640778