NameError: Cytotable not defined

[CesannevdM]

Hello there!
I have been trying Cytotable as a complete newbie to python. I have been following the tutorial for turning Cellprofiler CSVs into a Parquet file for Pycytominer in a command line.
I first created a virtual environment and installed the cytotable package there (which did succeed; i can see the package in the right folder).
Now i'm trying to import the package, which largely works except for the ForkProcess part. Later in the script, Python only seems to be able the read the cytotable command as a name, rather than as a command.

py

import os
import cytotable
Traceback (most recent call last):
File "", line 1, in
import cytotable
File "...venv\Lib\site-packages\cytotable_init_.py", line 24, in
from .convert import convert
File "...venv\Lib\site-packages\cytotable\convert.py", line 10, in
import parsl
File "...venv\Lib\site-packages\parsl_init_.py", line 21, in
from parsl.app.app import bash_app, join_app, python_app
File "...venv\Lib\site-packages\parsl\app\app.py", line 13, in
from parsl.dataflow.dflow import DataFlowKernel
File "...venv\Lib\site-packages\parsl\dataflow\dflow.py", line 27, in
from parsl.config import Config
File "...venv\Lib\site-packages\parsl\config.py", line 11, in
from parsl.executors.base import ParslExecutor
File "...venv\Lib\site-packages\parsl\executors_init_.py", line 1, in
from parsl.executors.flux.executor import FluxExecutor
File "...venv\Lib\site-packages\parsl\executors\flux\executor.py", line 21, in
from parsl.executors.base import ParslExecutor
File "...venv\Lib\site-packages\parsl\executors\base.py", line 13, in
from parsl.monitoring.radios.base import MonitoringRadioReceiver, RadioConfig
File "...venv\Lib\site-packages\parsl\monitoring_init_.py", line 1, in
from parsl.monitoring.monitoring import MonitoringHub
File "...venv\Lib\site-packages\parsl\monitoring\monitoring.py", line 14, in
from parsl.multiprocessing import (
...<4 lines>...
)
File "...venv\Lib\site-packages\parsl\multiprocessing.py", line 7, in
from multiprocessing.context import ForkProcess as ForkProcessType
ImportError: cannot import name 'ForkProcess' from 'multiprocessing.context' (C:\Users\User\AppData\Local\Python\pythoncore-3.14-64\Lib\multiprocessing\context.py)

SOURCE_PATH = os.environ["SOURCE_PATH"]
SOURCE_DATATYPE = os.environ["SOURCE_DATATYPE"]
DEST_PATH = os.environ["DEST_PATH"]
PRESET = os.environ["PRESET"]
CACHE_DIR = os.environ["CACHE_DIR"]

Result = cytotable.convert(
... source_path=SOURCE_PATH,
... source_datatype=SOURCE_DATATYPE,
... dest_path=DEST_PATH,
... dest_datatype="parquet",
... preset=PRESET,
... local_cache_dir=CACHE_DIR,
... # Reasonable chunking for large tables; adjust up/down if you hit memory limits
... chunk_size=30000,
... )
Traceback (most recent call last):
File "", line 1, in
Result = cytotable.convert(
^^^^^^^^^
NameError: name 'cytotable' is not defined

I have tried multiple names for the cytotable function (cytotable.py, convert, etc.). I have also tried adjusting directories, and updating everything, but none of it is working. Am I missing something that is causing the package to not load correctly?
Thanks in advance!

[d33bs]

Hi! I think you have installed CytoTable correctly. I believe the problem is that you’re using Python 3.14 on Windows, and one of CytoTable’s supporting packages does not currently work with that combination.

Because the first import cytotable step fails, Python cannot use cytotable.convert() afterward. That is why you see the second error, the function name itself is correct.

One solution is to install Python 3.13, create a new virtual environment, and install CytoTable there:

py -3.13 -m venv .venv
.\.venv\Scripts\Activate.ps1
python -m pip install --upgrade pip
python -m pip install cytotable

Then check that it loads:

python -c "import cytotable; print(cytotable.__version__)"

If that prints a version number, you can run your conversion again using cytotable.convert(...).

One small note: if you saved your own script as cytotable.py, rename it to something like run_conversion.py, since otherwise Python may confuse it with the installed package.

I'll work towards adding a test and support for Windows on Python 3.14 to help with future coverage there. In the meantime the workaround I mentioned should allow you to move forward. Let us know how it goes and thank you for using CytoTable!

[CesannevdM]

Hello! Thanks for the reply!
Unfortunately, i am still getting the same "ImportError: cannot import name 'ForkProcess' from 'multiprocessing.context'" error, even while using python version 3.13.
Do you happen to know any other things to try?
Thank you

[d33bs]

Hi! Thanks for testing 3.13 and reporting back - that's really helpful.

My first guess was only partly right, sorry about that. Looking again, this isn't actually about the Python version (3.13 vs 3.14 won't make a difference). The real issue is running CytoTable on native Windows.

Here's what's happening: your cytotable.convert(...) code is fine. The failure happens earlier, while Python is still importing CytoTable. CytoTable relies on a package called Parsl, and Parsl tries to import ForkProcess from Python's multiprocessing module. That ForkProcess only exists on Linux/macOS - it's never present on Windows, in any Python version. So the import fails, CytoTable never finishes loading, and that's why Python later says cytotable is "not defined."

One reliable fix might be to run CytoTable from a Linux-like environment. On Windows, the easiest way is WSL (Windows Subsystem for Linux). Once you have an Ubuntu/WSL terminal open:

python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install cytotable
python -c "import cytotable; print(cytotable.__version__)"

If that last line prints a version number, CytoTable loaded correctly and your conversion should run from that same environment.

Your CellProfiler CSVs on the Windows drive are still reachable from WSL through paths like:

/mnt/c/Users/YourUsername/...

(Minor side note: if you ever named your own script cytotable.py, rename it to something like run_conversion.py - but from your traceback that isn't the cause here.)

On our end, this points at a gap: we currently test on Linux and macOS but not Windows, and we don't surface this limitation clearly in the getting-started docs. I'll work on making it explicit. Thanks again for following up!