hwu

AUTHOR: Changshuai Wei

CONTACT: [email protected]

YEAR: 2016

LICENSE: Released under GNU General Public License, v2 (see COPYING.txt)

Precompiled excutable for windows and unix, as well as example data files can be found at http://changshuaiwei.github.io/software.html#HWU

COMPILATION: You will need a standard C++ compiler such as GNU g++ (version 3), as well as Boost c++ library (v1.55.0) and Eigen c++ library(v3.2).

Using HWU:

1:a simple one

hwu --bfile NIC --hwu --wIBS-wt --out NIC.rst

2:a complicated one

hwu --bfile NIC --hwu --dv 3 --cov all.cov.txt --flt --mp 20 --force-core --cov-wt cov1.txt --out NIC.rst

3:another complicated one

hwu --bfile NIC --hwu --screen 2 --wIBS-wt --mp 7 --PCwt 20 --out NIC.rst

##Options

--bfile string : read plink binary format data files

--file string: read text format data files

--hwu: perform HWU analysis

--out string: output files

--mp int: parallelize computations to int numbers of cpu cores

--wIBS-wt: use weighted IBS to construct latent population structure

--IBS-wt: use IBS to construct latent population structure

--cov-wt string : use covariates in string files to construct latent population structure

--dv int: output the most significant int SNPs (for high demensional scan)

--write-Wmt string : output kappa matrix for latent population structure to string file

--flt : use float accuracy when calculate the rank of SNPs (for high demensional scan)

--read-Wmt string: read kappa matrix stored in string file

--cov string: read covariates that used for covariates adjustment in string file

--PCwt int: calculate the first int eigen vectors for matrix kappa

--screen int: screen SNPs using standardized HWU, and calculate p-value of highest ranked int SNPs (for ultra-high demensional scan)

##compilation&build

First download Eigen and Boost under code/dependency. Unzip both, and then install boost by following steps here.

###For g++

For release: g++ -static -L dependency/boost_1_55_0/stage/lib -I dependency/eigen-3.4.0 -I dependency/boost_1_55_0 -pthread -msse2 *.cpp -lboost_system -lboost_thread -lrt -w -O3 -o build/hwu

For debug: g++ -static -L dependency/boost_1_55_0/stage/lib -I dependency/eigen-3.4.0 -I dependency/boost_1_55_0 -pthread -msse2 -g *.cpp -lboost_system -lboost_thread -lrt -w -O3 -o build/hwu_debug

###For VS code

Follow steps here. And refer to above g++ for args needed in corresponding json config.

###For visual studio (2015 version, outdated)

Make sure you compile with optimization enabled. This can easily gain you a factor of ten or more.

Configuring your project for optimal performance

Open the project properties page (Project | Properties), in the C/C++ folder open the Code Generation page. Under Enable Enhanced Instruction Set, switch to Streaming SIMD Extensions 2 (/arch:SSE2)
Open the project properties page (Project | Properties), in the C/C++ folder open the Language page. Under Open MP Support, switch to Yes (/openmp).
In case you are on a 64 bit machine, open the configuration manager (Project | Properties | Configuration Manager...). From here, select <New...> under Active solution platform:. Next, choose x64 and confirm the dialog with OK. Once you are back in the main window, switch from Win32 to x64 in order to create 64 bit builds of your program. 

Since boost::thread is used here, we might not use openMP for Eigen

need to build boost lib

use the VS comand prompt
run bootstrap.bat
run b2
go to properties->linker->general->additional library, add "$Root$\boost\boost_1_55_0\stage\lib"