TypeError: dtype '<class 'pandera.dtypes.Float128'>' not understood

[carbocation]

I installed these tools via $ pip install gwas-sumstats-tools on macOS. I tried to run gwas-ssf validate but got the following error:

$ gwas-ssf validate outfile.txt    

Traceback (most recent call last):
  File "/opt/homebrew/lib/python3.11/site-packages/pandera/engines/pandas_engine.py", line 160, in dtype
    return engine.Engine.dtype(cls, data_type)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/homebrew/lib/python3.11/site-packages/pandera/engines/engine.py", line 217, in dtype
    raise TypeError(
TypeError: Data type '<class 'pandera.dtypes.Float128'>' not understood by Engine.

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/opt/homebrew/bin/gwas-ssf", line 5, in <module>
    from gwas_sumstats_tools.cli import app
  File "/opt/homebrew/lib/python3.11/site-packages/gwas_sumstats_tools/cli.py", line 13, in <module>
    from gwas_sumstats_tools.validate import validate
  File "/opt/homebrew/lib/python3.11/site-packages/gwas_sumstats_tools/validate.py", line 7, in <module>
    from gwas_sumstats_tools.schema.data_table import SumStatsSchema
  File "/opt/homebrew/lib/python3.11/site-packages/gwas_sumstats_tools/schema/data_table.py", line 15, in <module>
    class SumStatsSchema:
  File "/opt/homebrew/lib/python3.11/site-packages/gwas_sumstats_tools/schema/data_table.py", line 41, in SumStatsSchema
    'p_value': Column(Float128, [
               ^^^^^^^^^^^^^^^^^^
  File "/opt/homebrew/lib/python3.11/site-packages/pandera/schema_components.py", line 88, in __init__
    super().__init__(
  File "/opt/homebrew/lib/python3.11/site-packages/pandera/schemas.py", line 1772, in __init__
    self.dtype = dtype  # type: ignore
    ^^^^^^^^^^
  File "/opt/homebrew/lib/python3.11/site-packages/pandera/schemas.py", line 1880, in dtype
    self._dtype = pandas_engine.Engine.dtype(value) if value else None
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/homebrew/lib/python3.11/site-packages/pandera/engines/pandas_engine.py", line 178, in dtype
    np_or_pd_dtype = pd.api.types.pandas_dtype(data_type)
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/homebrew/lib/python3.11/site-packages/pandas/core/dtypes/common.py", line 1702, in pandas_dtype
    raise TypeError(f"dtype '{dtype}' not understood")
TypeError: dtype '<class 'pandera.dtypes.Float128'>' not understood

[jdhayhurst]

Hi, please could you send a small sample of outfile.txt so that I can replicate the issue?

[EmilMiP]

has there been any development on this issue?

I am experiencing something very similar, except I am not even able to get gwas-ssf to print the help function.

I am commenting on this issue as I suspect it may be the same underlying issue.

If you need me to provide any information please let me know.

edit: I am on a windows system.

[privefl]

Same error here (just running gwas-ssf without any option or file). Using miniconda3 on Windows, with Python 3.11.4.

[ljwh2]

We will investigate whether we can replicate this locally. In the meantime could you try running in Docker?

[privefl]

I found it a bit difficult to use docker on Windows. Do you know any good tuto for it?

[ljwh2]

@privefl: https://docs.docker.com/desktop/install/windows-install/ Also some helpful tips here (from a different application): https://oboacademy.github.io/obook/howto/odk-setup/

[adrianICG]

Having the same issue here (also windows if that helps)

[ljwh2]

Apologies for the delay in fixing this issue while we await recruitment of a new staff member. In the meantime, if you are unable to validate your files please contact gwas-subs@ebi.ac.uk and we will be happy to support you further.

[EmilMiP]

I appreciate the offer and good luck with recruitment.

I have had to use a linux based system (the HPC cluster I work on) to validate the sumstats there, as I was unable to get it to work on my local (windows) machine. On the linux system, it work with any problems. However, I still hope the issue for windows can be solved in the future.