Building GLIMPSE
1. System requirements
GLIMPSE is a set of C++ tools covering the full process of imputation from low-coverage BAM files to imputed genotypes. In order to compile, we require a modern Linux operating system and a version of GCC > 4.4. We recommend to use the latest available version for your system.
For example running the following instruction on Ubuntu 20.04 focal:
sudo apt install build-essential
will install the GNU g++ compiler version 9.2. To check the version of your g++ compiler, simply run:
g++ --version
2. Required libraries
GLIMPSE requires two libraries installed on the system:
- HTSlib version >= 1.7: A C library for reading/writing high-throughput sequencing data.
- BOOST version >= 1.65: A set of peer-reviewed portable C++ source libraries. GLIMPSE uses two specific BOOST libraries: iostreams and program_options.
2.1 Building HTSlib
Building HTSlib is straightforward and does not require root privileges. Please refer to the HTSlib documentation for complete details. Here we provide a basic script to install HTSlib v1.11:
wget https://github.com/samtools/htslib/releases/download/1.11/htslib-1.11.tar.bz2
tar -xf htslib-1.11.tar.bz2
mv htslib-1.11 htslib
cd htslib
autoheader; autoconf; ./configure; #optional
make
After this, you'll find the libhts.a inside your current directory and the include files inside subdirectory: ./include/
2.2 Building Boost
As GLIMPSE only requires few of the boost libraries, we'll build the smallest possible boost build, without requiring root privileges. Please refer to the Boost installation instructions for complete details. Here we provide a basic script to the minimal build of Boost v1.73.0 required to run GLIMPSE:
wget https://boostorg.jfrog.io/artifactory/main/release/1.73.0/source/boost_1_73_0.tar.bz2
tar --bzip2 -xf boost_1_73_0.tar.bz2
cd boost_1_73_0
./bootstrap.sh --with-libraries=iostreams,program_options --prefix=../boost #where ../boost is your custom boost installation prefix
./b2 install
cd ../boost #change this to the folder you used as --prefix for the bootstrap script
After this, you will also find a copy of the Boost headers in the include/ subdirectory of the installation prefix (our current directory). The Boost static libraries (libboost_iostreams.a and libboost_program_options.a) are found in the subfolder ./lib of your installation prefix.
In the case there are problems finding the correct boost paths or with the boost installation, you can find additional instructions here: here.
2.3 Additional libraries
Make sure that the following standard library flags can be used by g++ on your system:
-lz,-lbz2and-llzmafor reading/writing compressed files.-lmfor basic math operations.-lpthreadfor multi-threading.
You can do so by checking the outcome of the following commands:
locate -b '\libz.so'locate -b '\libbz2.so'locate -b '\liblzma.so'locate -b '\libm.so'locate -b '\libpthread.so'locate -b '\libcurl.so'3. Compiling GLIMPSE
Download the last version of the GLIMPSE code using:
git clone https://github.com/odelaneau/GLIMPSE.git
Navigate to the downloaded folder using cd GLIMPSE.
You'll find there a folder containing all the software packages are other utility folders:
docs: source code of this website.maps: genetic maps for human taken from HapMap.tutorial: folder with the tutorial.README.md: front page of the GitHub.LICENSE: copy of the MIT license.makefileMakefile to compile all the programs at once, by recursively calling all the makefiles of the projects.
Each software in the suite contains the same folder structure:
bin: folder for the compiled binary.obj: folder with all binary objects.src: folder with source code.makefileMakefile to compile the program.
In order to compile a specific tool of the suite, for example GLIMPSE_phase, go in directory of the software (cd phase) and edit the specific makefile at lines 26-32 (desktop) so that the following variables are correctly set up (look at the paths already there for an example):
HTSSRC(line 26): path to the root of the HTSlib library, the prefix for HTSLIB_INC and HTSLIB_LIB paths,HTSLIB_INC(line 27): path to the HTSlib header files,HTSLIB_LIB(line 28): path to the static HTSlib library (filelibhts.a),BOOST_INC(line 29): path to the BOOST header files (often/usr/include),BOOST_LIB_IO(line 30): path to the static BOOST iostreams library (filelibboost_iostreams.a),BOOST_LIB_PO(line 31): path to the static BOOST program_options library (filelibboost_program_options.a),
If installed at the system level, static libraries (*.a files) can be located with this command:
locate libboost_program_options.a libboost_iostreams.a libhts.a
Once all paths correctly set up, proceed with the compilation using make. The binary can be found in the bin/ folder and will have a name similar to GLIMPSE_phase.
You will need to copy the modified makefile in each tool (folder) of the GLIMPSE toolsuite
3.1 Static binaries
In the case the installation is unsuccessful, static binaries (for Ubuntu x64 systems) of the latest release version are available in the static_bins folder. The binaries are just meant to make GLIMPSE available in systems where the dependencies are unavailable or difficult to install, but we strongly recommend to compile the working version and do not rely on static binaries, since binaries can suffer of a performance decrease.