koinapy Prosit TypeError: 'numpy.ndarray' object is not callable

[trvadams]

I am trying to use Prosit via koinapy to predict peptide CID fragments. I tried the sample python script under "Prosit_2020_intensity_CID": https://koina.wilhelmlab.org/docs#post-/Prosit_2020_intensity_TMT/infer

from koinapy import Koina
import numpy as np
import pandas as pd

# If you get a ModuleNotFound error install koinapy with `pip install koinapy`.
from koinapy import Koina
import numpy as np
import pandas as pd

# koinapy only takes the input it requires for the current model.
# if you want to compare multiple models you can use a dataframe wit all columns at the same time.
inputs = pd.DataFrame()  
inputs['peptide_sequences'] = np.array(["AAAAAKAK", "AAAAAKAK"]) 
inputs['precursor_charges'] = np.array([1,2]) 
inputs['collision_energies'] = np.array([25, 25])

# If you are unsure what inputs your model requires run `model.model_inputs`
model = Koina("Prosit_2019_intensity", "koina.wilhelmlab.org:443")
predictions = model.predict(inputs)

This returns the error message when calling model.predict(inputs):

TypeError: 'numpy.ndarray' object is not callable

I get the same problem when trying to use other models.

[LLautenbacher]

Thanks for trying out Koina. The code snippet you posted above is working for me. Can you share the full traceback that you see?

[trvadams]

Hi, yes the full traceback is here:

>>> predictions = model.predict(inputs)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/tadams/miniconda3/envs/ms/lib/python3.12/site-packages/koinapy/grpc.py", line 522, in predict
    return self.__predict_async(data, debug=debug)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/tadams/miniconda3/envs/ms/lib/python3.12/site-packages/koinapy/grpc.py", line 547, in __predict_async
    for i, data_batch in enumerate(self.__slice_dict(data, self.batchsize)):
  File "/home/tadams/miniconda3/envs/ms/lib/python3.12/site-packages/koinapy/grpc.py", line 330, in __slice_dict
    len_inputs = list(data.values())[0].shape[0]
                      ^^^^^^^^^^^^^
TypeError: 'numpy.ndarray' object is not callable

Python 3.12.4 Package versions: koinapy 0.0.1 numpy 1.26.4 pandas 2.2.2

[LLautenbacher]

I can see this error when I provide, a NumPy array as input instead of a Pandas Dataframe t o model.predict. Can you check the datatype and structure of inputs right before calling model.predict?

If that is the case I will improve error handling to display a more user-friendly error message.

[trvadams]

On my end, inputs appears to be a Pandas Dataframe:

>>> inputs
  peptide_sequences  precursor_charges  collision_energies
0          AAAAAKAK                  1                  25
1          AAAAAKAK                  2                  25
>>> inputs.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 2 entries, 0 to 1
Data columns (total 3 columns):
 #   Column              Non-Null Count  Dtype 
---  ------              --------------  ----- 
 0   peptide_sequences   2 non-null      object
 1   precursor_charges   2 non-null      int64 
 2   collision_energies  2 non-null      int64 
dtypes: int64(2), object(1)
memory usage: 180.0+ bytes

[LLautenbacher]

Dataframe support was added in version 0.0.2. Is there a reason you use version 0.0.1 instead of the latest 0.0.6?

[trvadams]

Aha, now I see the issue. I installed it via pip install koinapy yesterday, and now I see it seems like pip still defaults to 0.0.1:

pip install koinapy
Collecting koinapy
  Using cached koinapy-0.0.1-py3-none-any.whl.metadata (612 bytes)
Requirement already satisfied: tqdm<5.0.0,>=4.66.2 in /home/tadams/miniconda3/envs/ms/lib/python3.12/site-packages (from koinapy) (4.66.4)
Requirement already satisfied: tritonclient<3.0.0,>=2.42.0 in /home/tadams/miniconda3/envs/ms/lib/python3.12/site-packages (from koinapy) (2.46.0)
Requirement already satisfied: numpy>=1.19.1 in /home/tadams/miniconda3/envs/ms/lib/python3.12/site-packages (from tritonclient<3.0.0,>=2.42.0->koinapy) (1.26.4)
Requirement already satisfied: python-rapidjson>=0.9.1 in /home/tadams/miniconda3/envs/ms/lib/python3.12/site-packages (from tritonclient<3.0.0,>=2.42.0->koinapy) (1.17)
Requirement already satisfied: urllib3>=2.0.7 in /home/tadams/miniconda3/envs/ms/lib/python3.12/site-packages (from tritonclient<3.0.0,>=2.42.0->koinapy) (2.2.2)
Using cached koinapy-0.0.1-py3-none-any.whl (8.1 kB)
Installing collected packages: koinapy
Successfully installed koinapy-0.0.1

When specifying the latest version of koinapy, it doesn't recognize Python 3.12 as a valid version:

python -m pip install koinapy==0.0.6
ERROR: Ignored the following versions that require a different python version: 0.0.2 Requires-Python >=3.8,<3.11; 0.0.3 Requires-Python >=3.8,<3.11; 0.0.4 Requires-Python <3.11,>=3.8; 0.0.5 Requires-Python <3.11,>=3.8; 0.0.6 Requires-Python <3.11,>=3.8
ERROR: Could not find a version that satisfies the requirement koinapy==0.0.6 (from versions: 0.0.1)
ERROR: No matching distribution found for koinapy==0.0.6

I get the same error when trying to use the latest version of Python 3.11. However, when I use Python 3.10, pip installs 0.0.6 successfully:

python -m pip install koinapy                                                                                                                                        
Collecting koinapy                                                                                                                                                                                      
  Downloading koinapy-0.0.6-py3-none-any.whl.metadata (963 bytes)

The code snippet in the initial issue now works as expected:

>>> predictions = model.predict(inputs)
Prosit_2019_intensity:: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00,  1.98it/s

So it seems like it was a problem on my end with PyPi version control and newer versions of Python. Is this intentional, or could the PyPi be updated to support current Python versions?

[LLautenbacher]

Thanks for making me aware of this! This isn't intentional! I will look into this and update koinapy to be work with all current python version.

[trvadams]

Great, thanks for helping me figure it out. This tool is very cool, looking forward to incorporating it into some of my mass spec workflows 😎

[LLautenbacher]

This should be solved with the latest release for python versions up to 3.12.