make cuik-molmaker optional

[lucasattia]

Hello,
cuik-molmaker is currently required for fine-tuning KERMT. However, cuik-molmaker may not be compatible with many supercomputing environments, since it requires >= GLIBC_2.32. For example, NERSC perlmutter is compiled with GLIBC_2.31, so KERMT cannot be used on NERSC. Since cuik-molmaker mol graphs are equivalent to RDKit mol graphs, using cuik-molmaker should be made optional in molgraph.py.

[sveccham]

Hi @lucasattia,

Thanks for your interest in KERMT. KERMT can be used without cuik-molmaker but would require some code changes as you indicated. On the other hand, let me check if >= GLIB_2.32 is really required for cuik-molmaker.

[DonghaiZHAO-ZJU]

1. pip uninstall -y cuik_molmaker: Removes the previously installed, incompatible cuik_molmaker package.
2. apt-get update && apt-get install -y libeigen3-dev: Installs the system-level Eigen library development files, which are essential for compiling the C++ part of cuik-molmaker.
3. conda install ...: Installs necessary build dependencies within your kermt Conda environment. This includes pybind11 (for Python-C++ binding), librdkit-dev (RDKit development files), libboost-devel and
   libboost-python-devel (Boost library development files, potentially required by RDKit and pybind11), and crucially, gxx_linux-64=12 gcc_linux-64=12 to install the GCC 12 compiler suite, which fully supports
   the C++20 standard.
4. cd cuik-molmaker ...: Changes the directory to the cuik-molmaker source folder.
5. export ...: Sets environment variables like RDKIT_VERSION, PYTHON_VERSION, and CUIKMOLMAKER_CXX11_ABI, which are used by cuik-molmaker's setup.py for correct build configuration. Most importantly, export CXX=... and export CC=... explicitly instruct the build process to use the GCC 12 compilers installed in your Conda environment, overriding any older system-wide GCC, and resolving the C++ compiler
   standard mismatch.
6. rm -rf build: Cleans up any leftover artifacts from previous failed build attempts, ensuring a fresh compilation.
7. python setup.py build_ext --inplace: Compiles the cuik-molmaker C++ extension and places it in the current directory.
8. pip install .: Installs the cuik-molmaker package from the current source directory into your active Conda environment, making it importable by Python.
9. cd ..: Returns to the main project root directory.
10. export PYTHONPATH=$PWD && python scripts/save_features.py ...: Reruns your original command. With cuik_molmaker now correctly built and installed for your environment, the script should execute
    successfully.