In case someone is looking for an up-to-date way to install bcl2fastq2 (which is basically the relevant version of bcl2fastq): conda install is somewhat hidden, but

conda install bih-cubi::bcl2fastq2

works really well in early 2024.

(Seriously though, Illumina, is this a joke? those distributions are... something else. And no official Docker? Why?)

I took some notes last time I installed bcl2fastq and this worked for me on Ubuntu. However, it's an older version, so feel free to modify:

SOURCE_NAME=bcl2fastq2-v2.17.1.14.tar.zip
SOURCE_URL=ftp://webdata2:webdata2@ussd-ftp.illumina.com/downloads/software/bcl2fastq/$SOURCE_NAME
SOURCE_WORKDIR=/tmp/bcl2fastq

sudo apt-get update && \
    apt-get install -y build-essential libboost1.54-all-dev cmake zlibc libc6 unzip wget && \

$SOURCE_WORKDIR
cd $SOURCE_WORKDIR && \
    wget $SOURCE_URL 
cd $SOURCE_WORKDIR && unzip $SOURCE_NAME && \
    tar xzf $(basename $SOURCE_NAME .zip).gz && \
    mkdir $SOURCE_WORKDIR/bcl2fastq/build && \
    cd $SOURCE_WORKDIR/bcl2fastq/build && \
    ../src/configure --prefix=/usr/local && \
    make && \
    make install

Hi

I generated deb packages and zip files for binaries for the latest 2 versions of bcl2fastq2. Feel free to try these.

All compiled on ubuntu with static library linking so there should not be any resource dependencies. Zip packages can be used on any linux distro just by adding the folder to the PATH. I tried these on Manjaro and Ubuntu so far and no issues detected. Debian packages are not converts from rpm. Fresh compiled from source code under ubuntu 18.04.

Bcl2fastq 2.19 deb package compiled via gcc 7.3

Bcl2fastq 2.20 deb package compiled via gcc 7.3

Bcl2fastq 2.20 zip package compiled via gcc 8.2

Bcl2fastq 2.20 zip package compiled via gcc 7.3

Sorry about that. I am a long time registered user in GATK forums my nickname is SkyWarrior there. I have been in bioinformatics for about 5 years and before that I was all a wetlab guy. As a hobbyist I dealt with FOSS for a long time and after I quit wetlab business as a researcher I decided to give my linux and math skills a go for bioinformatics and it has paid pretty well.

Again I am sorry if any of these links felt like a scam but believe me it is not. I am publishing the steps that I did to generate these deb files so anyone can make their of if they don't want to download but still need a private repository for themselves to keep a copy of working software.

Here are the things you need to get things up and running for this compilation to go successfully. Internet links are full of recommendations of how to get this thing done on older versions of ubuntu or other linux distro but I tried this on the latest stable and rolling ubuntu.

1- Get Ubuntu 18.04 or 18.10 running on a computer or a VM. I use VM or docker image because it is much more handy and I can keep that for later versions to build if needed. 2- Run the following commands to get your compilers and libs up to date

sudo apt-get update
sudo apt-get install gcc build-essentials libbz2-dev zlib1g-dev

These commands will install gcc 7.3 in 18.04 and gcc 8.2 in 18.10. Once these are done successfully follow the steps below.

Steps to create these deb packages your own.

• Download the latest source code from illumina website and extract the bcl2fastq folder inside another empty folder that would keep your work tidy.

• Once you extracted bcl2fastq folder create a folder named build next to bcl2fastq folder.

• Now open up a terminal window inside the build folder and type the following command that is needed for proper libboost-1.54 compilation from the source code provided.

  export CPLUS_INCLUDE_PATH=/usr/include/x86_64-linux-gnu:$CPLUS_INCLUDE_PATH

This command restores the stat.h file for a missing header problem in libboost compilation.

• Now run the below command to start the configuration.

  ../bcl2fastq/src/configure --static --prefix=/usr/local --build-type Release --package-type=deb

• This command will go through the steps of creating static library links required to build bcl2fastq in the following step.

• Once this command is completed successfully it is time to modify a file that prevents deb package to be generated with the proper version info successfully. Inside the build folder there is a file named CPackConfig.cmake. Open this file with a text editor and change the following line (version is for the 2.20 but you get the point.)

  SET(CPACK_PACKAGE_VERSION "v2.20.0.422") to SET(CPACK_PACKAGE_VERSION "2.20.0.422")

• After that run the following command inside the build folder again as usual.

  make package

After these steps you should have a deb package generated successfully inside the folder build. You can use that package to install to any debian based distro.

There is also another folder named _CPack*. Inside that folder you will also find compiled binaries that can be used in any linux distro that is not compatible with deb packages.

It is sad that Illumina does not provide this by default but it is nothing impossible with a few tricks explained here.

Here are the screenshots of the running package on my Windows 10 Desktop with Ubuntu on Windows feature. It just works

I am preparing a video link for everyone to follow what I did visually here. Just give me an hour.

Hi

Sorry that this is a very late reply for a repository for this build. Finally I generated a docker image for this build.

It can be had from skysbiodocker/bcl2fastq2

docker pull skysbiodocker/bcl2fastq2

There is another docker image that I used to use which is the illumina rpm build based on centos6.9 however that image is incompatible with later kernel versions especially 4.19. I also noticed that illumina rpm build of the 2.20 tool is also segfaulting under latest kernels but If you wish to use that that is also available

docker pull skysbiodocker/bcl2fastq

Beware that kernel incompatibilities may kill this image.

Here is a build on CentOS 8 using the 8.3.1 version of the gcc tools. The patch gets rid of all the warnings in the build of bcl2fastq itself, and the binary runs. It has not been tested extensively, unknown if the results are correct. Note however that the configure step is still full of much strangeness, including:

1.  building its own cmake
2.  finding a more recent libxslt but building its own older version.
3.  finding a more recent libxml2 but building its own older version
4.  scads of compiler warnings

The build target uses an Lmod module at the end. Change TOPDIR and skip the module generate if you do not need that.

  #2020/04/24 on CentOS 8
  pversion=2.20.0.422
  TOPDIR=/usr/common/modules/el8/x86_64/software/bcl2fastq/${pversion}-CentOS-vanilla
  #previous is where it will be installed
  cd /usr/common/src #work directory only , /tmp would be OK
  wget ftp://webdata2:webdata2@ussd-ftp.illumina.com/downloads/software/bcl2fastq/bcl2fastq2-v2-20-0-tar.zip
  unzip bcl2fastq2-v2-20-0-tar.zip
  /bin/rm bcl2fastq2-v2-20-0-tar.zip
  gunzip -c bcl2fastq2-v2.20.0.422-Source.tar.gz  | tar -xf - 
  /bin/rm bcl2fastq2-v2.20.0.422-Source.tar.gz
  cd bcl2fastq/src
  wget https://saf.bio.caltech.edu/pub/software/molbio/bcl2fastq_2.20.patch
  cat bcl2fastq_2.20.patch | patch -p0
  export BOOST_INCLUDEDIR=/usr/include/boost169
  export BOOST_LIBRARYDIR=/usr/lib64/boost169
  mkdir build
  cd build
  ../configure --prefix=$TOPDIR --with-unit-tests \
   --build-type=Release --verbose \
  2>&1 \
   | tee configure_2020_04_24.log
  make 2>&1 | tee build_2020_04_24.log
  make install 2>&1 | tee install_2020_04_24.log
  #If you use Lmod this script is handy.  Modify to fit your site:
  #  https://saf.bio.caltech.edu/pub/software/linux_or_unix_tools/module_generate_from_directory.sh
  module_generate_from_directory.sh \
     bcl2fastq \
     $pversion\
     CentOS/vanilla \
     $TOPDIR \
     "Demultiplex sequencing data and convert base call (BCL) files into FASTQ files." \
     "https://support.illumina.com/sequencing/sequencing_software/bcl2fastq-conversion-software.html"