theislab
`pertpy` Installation Fails on Google Colab Due to `numpy` AttributeError
Description
I tried installing pertpy on Google Colab using the following commands:
!pip install pertpy
!pip install pertpy[coda]
import pertpy as pt
However, when attempting to import pertpy, I encountered the following error:
AttributeError Traceback (most recent call last)
<ipython-input-3-5d616454759d> in <cell line: 0>()
2 get_ipython().system('pip install pertpy[coda]')
3
----> 4 import pertpy as pt
5 get_ipython().system('pip install scanpy')
6 get_ipython().system('pip install plotnine')
24 frames
/usr/local/lib/python3.11/dist-packages/numpy/__init__.py in __getattr__(attr)
322 warnings.filterwarnings("ignore", message="numpy.dtype size changed")
323 warnings.filterwarnings("ignore", message="numpy.ufunc size changed")
--> 324 warnings.filterwarnings("ignore", message="numpy.ndarray size changed")
325
326 def __getattr__(attr):
AttributeError: module 'numpy' has no attribute 'bool'.
np.bool was a deprecated alias for the builtin bool. To avoid this error in existing code, use bool by itself. Doing this will not modify any behavior and is safe. If you specifically wanted the numpy scalar type, use np.bool_ here.
The alias was originally deprecated in NumPy 1.20; for more details and guidance see the original release note at:
https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations
⚠️ THIS ISSUE PREVENTS PERTPY FROM BEING USED IN GOOGLE COLAB. ⚠️
Steps to Reproduce
1.Open a new Google Colab notebook, start a v2-8 TPU runtime. 2.Run the following commands:
!pip install pertpy
!pip install pertpy[coda]
import pertpy as pt
3.Observe the AttributeError related to numpy.bool.
Dear @LCGaoZzz ,
I'm sorry for the inconveniences. The issue is that the scipy version on google colab is not recent enough if you install pertpy. If you run pip install -U scipy after having installed pertpy, this issue is mitigated. You might run into another compatibiliy issue related to import hooks. See https://github.com/googlecolab/colabtools/issues/5000. This out of my control.
This is how I got it to work:
!pip install pertpy[coda]
!pip install -U scipy
!pip install -U torch
import sys
import types
from importlib.machinery import ModuleSpec
# Find and fix ANY import hook missing find_spec
hooked_imports = []
for i, finder in enumerate(sys.meta_path):
# Skip built-in finders that already have find_spec
if hasattr(finder, 'find_spec'):
continue
# Only fix hooks that have find_module
if hasattr(finder, 'find_module'):
hook_name = finder.__class__.__name__ if hasattr(finder, '__class__') else str(finder)
hooked_imports.append(hook_name)
def find_spec_wrapper(self, fullname, path, target=None):
loader = self.find_module(fullname, path)
if loader is not None:
return ModuleSpec(name=fullname, loader=loader)
return None
# Bind the method to the instance
finder.find_spec = types.MethodType(find_spec_wrapper, finder)
print(f"Fixed {len(hooked_imports)} import hooks: {', '.join(hooked_imports)}")
# Now try importing sympy directly
try:
import sympy
print("Successfully imported sympy")
# Then try pertpy
import pertpy as pt
print("Successfully imported pertpy")
except Exception as e:
print(f"Error: {e}")
Unfortunately, colab is a mess. I'll try to set better minimum boundaries soon to help mitigate such issues but I also partially blame Colab for this.
Hope this helps!
