Set upper bound for numpy version

[a-r-j]

Hiya,

I run into an error with newer numpy versions (1.24.0):

File ~/work/graphein/graphein/graphein/molecule/features/graph/molecule.py:67, in <dictcomp>(.0)
     62     descriptors = {
     63         k: v for k, v in descriptors.items() if k in descriptor_list
     64     }
     66 # Compute descriptors
---> 67 desc = {d: descriptors[d](mol) for d in descriptors}
     69 # Process Outformat
     70 if return_array:

File /usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/rdkit/Chem/GraphDescriptors.py:124, in Ipc(mol, avg, dMat, forceDMat)
    121       mol._adjMat = dMat
    123 adjMat = numpy.equal(dMat, 1)
--> 124 cPoly = abs(Graphs.CharacteristicPolynomial(mol,adjMat))
    125 if avg:
    126   return entropy.InfoEntropy(cPoly)

File /usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/rdkit/Chem/Graphs.py:43, in CharacteristicPolynomial(mol, mat)
     41 I = 1. * numpy.identity(nAtoms)
     42 An = A
---> 43 res = numpy.zeros(nAtoms + 1, numpy.float)
     44 res[0] = 1.0
     45 for n in range(1, nAtoms + 1):

File /usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/numpy/__init__.py:284, in __getattr__(attr)
    2[81](https://github.com/a-r-j/graphein/actions/runs/3680239676/jobs/6320741907#step:8:82)     from .testing import Tester
    2[82](https://github.com/a-r-j/graphein/actions/runs/3680239676/jobs/6320741907#step:8:83)     return Tester
--> 284 raise AttributeError("module {!r} has no attribute "
    285                      "{!r}".format(__name__, attr))

AttributeError: module 'numpy' has no attribute 'float'

From this article it appears some aliases (like np.float) have been deprecated. As a result, I think the supported NumPy versions should be constrained or the underlying code refactored to support new versions.

[kuelumbus]

Hi. This was already fixed in the RDKit repository (rdkit/rdkit#5897). The next RDKit PyPI release should work Numpy >=1.24. In the meantime, please use an older numpy version. Hope this works for you.

Please leave this issue open until the next version is released.

[kuelumbus]

Should be fixed in the 2022.9.4 release.