This is a faster implementation of https://github.com/atifrahman/HAWK/blob/master/log_reg_case.R and https://github.com/atifrahman/HAWK/blob/master/log_reg_control.R. This work is done as course project of CSE6411 in April 2018 semester. This code are part of the implementation of method described in this paper https://elifesciences.org/articles/32920.
boost library https://www.boost.org/
To build the executable you need to have make tool. You have to edit the Makefile to give the location of boost library include file ($BOOST_HOME) and boost library location ($BOOST_LIB).
make
The executable has the capability to divide the work among multiple worker. This can lead to potential speedup in multi core platform. The executable calling format is :
./log_reg_case.out [-t number of worker] [-c covariate file name] [-p number of principle components]
./log_reg_control.out [-t number of worker] [-c covariate file name] [-p number of principle components]
For example, to run the code with 2 worker, 3 principle components, and has a covariate file, "cov_0.txt"
./log_reg_case.out -t 2 -p 3 -c cov_0.txt
./log_reg_control.out -t 2 -p 3 -c cov_0.txt
To run with single worker calling the executable without any argument is enough. Passing -t 1 will do the same
Test folder contains some dummy data (See https://github.com/atifrahman/HAWK and paper for details of data). To run the code (both log_reg_case.out and log_reg_control.out) from base dir,
cd test
../log_reg_case.out > pvals_case_top.txt
../log_reg_control.out > pvals_control_top.txt
Each executable dumps the p value in the redirected files. (In order to run convertToFasta.cpp, you must have the files: pvals_case_top.txt and pvals_control_top.txt)
To compare the quality of output with the original R code output you can use r_cpp_output_compare.py by adding the output containing file name (both R and cpp output file). The comparison is shown by evaluating mean square error and std. dev. of all the pvalue and corresponding R output's p values.