I'm having some problems with the output of the predictions

[lennyliu2000]

Hello, I am sorry to bother you again after such a long time, but recently I have been using your SimSpread algorithm for peptide target prediction, and the data I used are peptide-target interaction matrix and peptide-peptide similarity matrix, and I have followed the algorithm code you gave me, but in the end I can't get any valid predictions, all the prediction results are 0! Even the relationships that are already there cannot be predicted. I would like to ask you what caused this situation, is it the problem of the data I prepared or the problem of the algorithm, I hope you can give me some advice and teach me the way to solve this problem, I appreciate it!
I've also attached an example of the predictions obtained by running the code, which you can also see, which really doesn't contain any valid predictions
(P.S. I'm not using the native Julia code you provided here, but the code I wrote in Python according to the principle of your algorithm, which is basically the same as yours, so there's no problem with my code in terms of syntax and logic)
Predictions_0.20.csv

[cvigilv]

Hi! No problem, I'm here to help.

Since you are not using the version provided here, it's difficult to provide guidance to what could be happening so I'm going to suggest the following:

1. Did you reproduce the papers results with your python version?
2. Empty prediction (score = 0) mean the algorithm didn't find anything similar to what you are searching. This is intended. Have you done a parametrization of the similarity cutoff? Maybe it's to high and it results in this behavior.
3. You could use juliacall and python.jl to run the Julia code inside python. Check that if you can

Let me know so I can proceed helping you

[lennyliu2000]

Thanks for your suggestions, here's the deal:
First of all, I did not use the data in your paper to reproduce the results in your paper, but directly used my data to try to make predictions.
Second, I did not change the cutoff parameter, but used the optimal value of 0.2 that you set in your paper.
Thirdly, I have checked this question elsewhere, and concluded that it may be related to the data I used, and that the interaction matrix in my data may be too sparse to allow accurate resource diffusion.
Finally, do you have any background knowledge in Python? If you can, could you help me look at the python code I have written? I don't mean anything else, I just want to make sure that there are no other problems with my code, after all, you are a great expert in this field, I am honored to have your guidance, thank you!

[cvigilv]

Try using a different cutoff, since that cutoff was selected specifically for the combination of parameters we had in our problem.

I would recommend on first reproducing the results of the paper and later using the method for your use cases since it's the only way tu ensure the Python version works the same as the Julia one.

You may add the python code here, but as I said before, I would recommend using JuliaCall to use the SimSpread library directly in python.

[lennyliu2000]

Ok, thanks for your answer, I will make new adjustments and try the algorithm, this is my python code, you can take a look at it, if there is any problem you just point it out!
nbi.txt
SimSpread_PP.txt