Installing SINA

You can install SINA

  1. using Bioconda (recommended)
  2. from pre-compiled tarballs (alternate)
  3. or build SINA from source (for developers)

Install using Bioconda

SINA is available as a Conda package in the Bioconda channel. Check the package info page for more information.

To install, follow these steps:

  1. Install Miniconda (skip if you’ve got conda already)

    Download the Miniconda installer (links for MacOS and Linux), execute it and follow the instructions it shows in the shell:

    # if you are on MacOS
wget https://repo.continuum.io/miniconda/Miniconda3-latest-MacOSX-x86_64.sh
# if you are on Linux
wget https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh

# then
sh Miniconda3-lastest-*-x86_64.sh

  2. Add the Conda-Forge and Bioconda channels:

    conda config --add channels defaults
conda config --add channels bioconda
conda config --add channels conda-forge

  3. Install SINA into a conda environment:

    conda create -n sina sina

  4. Activate and use environment:

    conda activate sina
sina --help

SINA should also work fine if installed with conda into the base environment. If you encounter problems, try the separate environment.

Install from pre-compiled tarballs

Tar archives containing pre-compiled binaries and requisite libraries are available on the SINA releases page at Github. Head on over there and download the Linux or Macos one. Inside the folder created by unpacking the archive, you should find a sina executable:

tar xf ~/Downloads/sina-1.7.2-linux.tar.gz
~/Downloads/sina-1.7.2-linux/sina --help

To install SINA system wide, place the contents of the archive in /opt and create symlinks into /usr/local/bin:

wget https://github.com/epruesse/SINA/releases/download/v1.7.2/sina-1.5.0-linux.tar.gz
sudo tar xf sina-1.7.2-linux.tar.gz -C /opt
rm sina-1.7.2-linux.tar.gz
sudo ln -s /opt/sina-1.7.2-linux /opt/sina
sudo ln -s /opt/sina/bin/sina /usr/local/bin/sina

Build from source code

Building SINA from source can be challenging because SINA depends on the ARB development libraries. Pre-compiled versions of these are currently only available from Bioconda, so you may have to start by building ARB from source.

Prerequisites

When building from a source tar ball from the SINA releases page, you will need to install:

  • ARB >= 6.0.0
  • Boost >= 1.62 - Boost Thread - Boost Program Options - Boost IO Streams - Boost Filesystem - Boost Serialization (SINA < 1.5) - Boost System - Boost Unit Test Framework (when building / running tests)
  • TBB >= 2017
  • zlib

When building from the raw git source code directly, you will additionally need:

  • Autoconf
  • Automake
  • Libtool
  • pkg-config

To build the documentation, you need:

  • Sphinx >= 1.8

The easiest way to get all requirements is using Conda:

if test $(uname) == Linux; then
  dist_extra="gcc patchelf coreutils"
else
  dist_extra="llvm"
fi
conda create -n sina_build automake autoconf libtool pkg-config boost arb-bio-devel \
                git tbb tbb-devel glib libiconv bc sed sphinx $dist_extra

Building

  1. If you are building from git, start by checking out the source code and generating the configure script:

    git clone https://github.com/epruesse/SINA.git sina
cd sina
autoreconf --force --install

  2. Run the configure script, pointing it at all required libraries as necessary and choosing features and build types:

    ./configure --prefix=install_location \
            --with-arbhome=path_to_arbhome \
            --with-boost=path_to_boost_install \
            --with-boost-libdir=path_to_boost_libs

    If you used conda to install your dependencies, this line should work:

    conda activate sina_build
mkdir build
cd build
../configure --prefix `pwd`/install \
             --disable-docs \
             --with-tbb=$CONDA_PREFIX \
             --with-boost=$CONDA_PREFIX \
             --with-boost-libdir=$CONDA_PREFIX/lib \
             --with-arbhome=$CONDA_PREFIX/lib/arb \
             LDFLAGS="$LDFLAGS -Wl,-rpath,$CONDA_PREFIX/lib"

    Essential options to configure:

    --prefix=PATH (/usr/local)

    Set the folder under which ./bin/sina, ./lib/libsina.so (or .dylib), etc. will be installed.

    --with-tbb=PATH

    Set the location of libraries and headers for the Intel Threading Building Blocks library.

    --with-boost=PATH

    Set the location of the boost header files (without the include/ part).

    --with-boost-libdir=PATH

    Set the location of the boost lib folder. Often, this is the value you used for --with-boost with /lib appended.

    --with-arbhome=PATH

    Set the location of the ARB build directory. Not needed if you have $ARBHOME set to point to the place you built ARB. When using the Bioconda package arb-bio-devel, use $CONDA_PREFOX/lib/arb where $CONDA_PREFIX is the root of the environment you installed ARB into.

    --with-buildinfo=TEXT

    Set an additional string to be added to the version to identify your build.

    --enable-code-coverage

    Add compiler flags to collect code coverage statistics.

    --enable-debug

    Enable debug options (sets -DDEBUG -O0 -ggdb3 instead of -DNDEBUG -O2 -g).

    --enable-asan

    Enable address sanitizer (sets -fsanitize=address).

    --enable-fat-tar

    Alters the build so that make bindist-gzip constructs a fully contained tar archive of the build.

    --enable-profiling

    Add compiler flags collecting profiling statistics (-pg).

    --disable-docs

    Do not build the documentation.

    If you installed the dependencies in system wide, standard FHS locations, the configure script should detect the locations correctly. Otherwise you may have to use the --with-something options to point it at the right places. If things go wrong, the full error messages will be in config.log.

  3. Build SINA (replace <number of cpus with however many cores you’ve got):

    make -j<number of cpus>
make install

    To build binary archives (see also --enable-fat-tar), use:

    make bindist-gzip2

    To run unit tests, call:

    make check

    To run only part of the tests, call:

    make check-filtered P=pattern

    where pattern matches the name(s) of the test you wish to (re)run.

    To run unit tests collecting code coverage, call:

    make check-code-coverage

    To see the full command line of compiler and linker instead of the abbreviated display, append V=1 to the make commandline.