The QtPlatz (as qtplatz) is the software foundation for chromatography and mass spectrometry instrument control and data systems. It is intended to implement molecular targeting applications based on accurate masses. The software is written in C++ and can run on Windows 10/11, Mac OS 12 and onward, and Debian Linux and its variants, including Raspberry Pi.

Visit http://www.ms-cheminfo.com

The executable binary file is on the github.com release. On the recent version of macOS, it may raise an error to open qtplatz.app ion as “qtplatz.app” is damaged and can’t be opened. You should eject the disk image or “qtplatz”は壊れているため開けません。 ディスクイメージを取り出す必要があります on the Japanese environment. This error could be resolved by issuing the following command from the Terminal.

xattr -d com.apple.quarantine ~/Desktop/qtplatz.app

It might be possible to raise an error 'XKB::XKB not found' when configuring the cmake project. In such a case, check 'libxkbcommon-dev' is installed.

sudo apt install libxkbcommon-dev

1. git

2. cmake 3.18.2

3. c++17 compliant c/c++ compiler (Xcode 12|gcc-6.3|msvc-16)

4. Qt 5.15.2 | Qt 6.5.3 (download)

5. QWT 6.2

6. Boost-1.83 (also work with 1.73 and later versions)

7. RDKit 2023.03 (optional, but recommended) -- RDKit should be built with a version of boost, which matches the qtplatz build.

8. OpenCV 4.5 (optional)

9. Python3 (optional)

10. netcdf-c (optional; ENABLE_FILTER_BLOSC and ENABLE_FILTER_ZSTD should be disabled or get killed:9 on mac-bundle)

    QtPlatz uses boost_serialization for binary data storing on file. A file serialized by an older version of boost can be opened in a newer version of boost, but no reverse compatibility. To check the version of boost_serialization, see the BOOST_ARCHIVE_VERSION macro value located in the file boost-source-dir/libs/serialization/src/basic_archive.cpp.

• Xcode 12 (command-line tools) needs to be installed for macOS.
• Visual Studio 2019 needs to be installed for Windows.
• Qt 5.12 or later version needs to be installed for the target OS.
  • Qt 5.12.2 does not work on Debian based linux -- use Qt 5.12.1 instead.
  • Qt 6 is not supported yet (may it work, though).

1. Xcode Version 12.3 (12C33)
2. Download the respective qt5 installer from download page, and follow the instructions.

Go to the <qtplatz-source-dir>/scripts directory; there are script files for installing dependency software modules. Assume the qtplatz source is stored under ~/src

cd ~/src/qtplarz/scripts
make dependency
make boost
make qwt
make rdkit

• Boost 1.75 (1.79 preferred) is required for the needs of the boost::json library.
• Qwt 6.2.0 is required since QtPlatz Version 5.2.11 and onward.

cd ~/src/qtplatz
./bootstrap.sh
cd ~/src/build-Darwin-i386/qtplatz.release # <- build-Linux-x86_64 for Linux
make

QtPlatz binary to be built under the ~src/build-Darwin-i386/bin (~/src/build-Linux-x86_64/bin) directory. Installation for Linux is essentially the same step with macOS.

First of all, if you got a new M1 Mac computer migrated from your old Intel Mac, it may contain x86_64 binaries under /usr/local, etc. Especially installed libraries for QtPlatz build, such as boost, rdkit, and dependencies, will be detected by the cmake configuration phase but may fail at the link phase because arm64 and x86_64 binary cannot link. Fortunately, homebrew on M1 Mac installs all files under /opt/homebrew/ instead of /usr/local/, which is less problematic. However, we also need to use MacPort to install maeparser (rdkit depends on it) and llvm.

1. Xcode Version 13.2.1 13C100
2. Download the qt5 source code (either download from download or git.

This step is straightforward but expects one error during the compiling phase.

$ cd ~/src
$ git clone git://code.qt.io/qt/qt5.git --branch 5.12
$ cd qt5
$ ./init-repository --module-subset=qtbase
$ git checkout 5.15.2
$ mkdir ~/src/build-Darwin-arm64/qt5-build
$ cd ~/src/build-Darwin-arm64/qt5-build
$ ~/src/qt5/configure -prefix /opt/Qt/5.15.2 QMAKE_APPLE_DEVICE_ARCHS=arm64 -opensource -confirm-license
$ make -j10

The above command set may hit a compile error described here In such a case, a quick fix is to apply qtplatz/scripts/qt5-5.15.2.patch (or simply add #include #include <CoreGraphics/CGColorSpace.h> int src/plugins/platforms/cocoa/qiosurfacegraphicsbuffer.h file. You should be able to build qt5 development files after running 'make' and 'make install' commands.

qwt-6.2.0 can be installed as follows;

/opt/Qt/5.15.2/bin/qmake -r qwt.pro QMAKE_APPLE_DEVICE_ARCHS=arm64
make
sudo make install

The rest of the dependent modules can be installed as follows:

cd ~/src/qtplarz/scripts
make dependency
make boost
port install maeparser
port install llvm
make rdkit

Recently released Xcode 15 contains Clang-15 toolchains that break build for at least boost-1_79 through boost-1.82 due to c++-17 removing std::unary_function and std::binary_function, which were marked as deprecated since c++11. See boostorg/container_hash#22 A quick fix for this is to modify two header files, as listed below.

// boost/functional.hpp line 27

        namespace detail {
#if defined(_HAS_AUTO_PTR_ETC) && !_HAS_AUTO_PTR_ETC || (BOOST_CXX_VERSION >= 201703)
            // std::unary_function and std::binary_function were both removed
            // in C++17.

// boost/container_hash/include/boost/container_hash/hash.hpp

#if defined(BOOST_NO_CXX98_FUNCTION_BASE) || (BOOST_CXX_VERSION >= 201703)
        template <typename T>
        struct hash_base
        {
            typedef T argument_type;
            typedef std::size_t result_type;
        };
#else
        template <typename T>
        struct hash_base : std::unary_function<T, std::size_t> {};
#endif

boost-1_83 is a recommended boost version with QtPlatz 5.6.1. An issue previously reported below was resolved. It happened due to the boost::json change on boost-1.81, where they re-design conversion classes -- it requires application code change, and it has been done as of QtPlatz 5.6.1. As of 20th Sep. 2023, I have tested with boost-1_83, which seems to have no above issues and is able to be built for boost libraries, maeparser, and RDKit. However, the compile error was raised in 'boost/json/value.hpp', which uses many of the classes in qtplatz.

Although already reported on the web pages, execution of qmake command raises the following error so that no generator makefile is generated properly. The error message isn't clear about the meaning of search paths, whether command search path or library search path, though it seems that it is a dynamic library search path according to old web articles.

Project ERROR: failed to parse default search paths from compiler output

This problem isn't an issue for the qtplatz build; however, it is an issue for building qwt library.

1. Visual Studio 2022
2. WiX toolset
3. Python 3.7 (optional)

Tool for dll dependency diagnostics: https://github.com/adamrehn/dll-diagnostics

cd %USERPROFILE%\src\qtplatz\scripts
.\windows-bootstrap.bat
nmake boost
nmake eigen
nmake maeparser
nmake rdkit
nmake opencv

cd %USERPROFILE%\src\qtplatz
.\bootstrap.bat package
nmake package

To support ANDI/Chromatograpy (.cdf) data read into QtPlatz, it requires netcdf.dll installed on Windows. Building and installing netcDF on macOS and Linux is straightforward; in contrast, building the netCDF library on Windows is problematic. Although there are prebuilt binary exist on https://downloads.unidata.ucar.edu/netcdf/, there is a bug in the CMakeTarget.cmake file, where the dependent libraries for libz.lib and libcurl.lib were under hard-coded full path names, such as below:

set_target_properties(netCDF::netcdf PROPERTIES
  INTERFACE_INCLUDE_DIRECTORIES "${_IMPORT_PREFIX}/include;${_IMPORT_PREFIX}/include"
  INTERFACE_LINK_LIBRARIES "hdf5-shared;hdf5_hl-shared;C:/share/VS15/x64/lib/zlib.lib;C:/share/VS15/x64/lib/libcurl_imp.lib"
)

The INTERFACE_LINK_LIBRARIES shown above should be fixed as:

  INTERFACE_LINK_LIBRARIES "hdf5-shared;hdf5_hl-shared;${_IMPORT_PREFIX}/lib/zlib.lib;${_IMPORT_PREFIX}/lib/libcurl

The binary package comes with required import libs and DLLs, so it should remove the full path name from INTERFACE_LINK_LIBRARIES.

I have tried to build netCDF from source, depending on the libraries libz, libcurl, and libhdf5. All those dependent libraries can be installed on Windows straightforwardly. However, netCDF (CMake) cannot successfully be configured. Even if it succeeded in configuring, the build step on VS2022 failed for the netcdf3 project step. Finally, I gave it up.