some questions

[legendaryInnovation]

how to solve it

[njmarko]

I will check what arguments I used, and I will respond here.

Also, I will try to update the repo as soon as I have enough time with that kind of information. I will also add deployment for this later.

[legendaryInnovation]

I will check what arguments I used, and I will respond here.

Also, I will try to update the repo as soon as I have enough time with that kind of information. I will also add deployment for this later.

Thank you

[njmarko]

The code for handling the dataset is from https://github.com/andreeadeac22/graph_coattention
I will list the steps that worked for me on Windows 10 with PowerShell.
I will add requirements.txt for this repo soon, so you can install them.

The first step is to download the dataset into the /data/decagon folder.
cd .\data\
mkdir decagon
cd .\decagon\

Download the dataset with wget (if it works, it has some problems if you are using Windows)
wget http://snap.stanford.edu/decagon/bio-decagon-combo.tar.gz;

Or try to use curl
curl http://snap.stanford.edu/decagon/bio-decagon-combo.tar.gz

If this doesn't work, just download the dataset manually by visiting the link http://snap.stanford.edu/decagon/bio-decagon-combo.tar.gz and chose the folder /data/decagon/ for the download location.

The next step is to unpack the data with tar
tar -xvzf bio-decagon-combo.tar.gz;

After that, you can run the data_download.py script with the following arguments
decagon -p decagon/

The progress bar doesn't seem to work in this case. Just wait until the script finishes executing. It will probably take around 10 minutes.

[njmarko]

The next step is to run data_preprocess.py with these arguments
decagon -p ./

[njmarko]

Finally, run split_cv_data with these arguments
decagon -p ./

[njmarko]

Executing split_cv_data.py takes around 20 to 30 minutes.
I added requirements.txt to the repo for now.
Also, there should be a folds directory in /data/decagon. I added a fix for that now, but you can also create it manually.

Let me know if these steps worked for you.

[legendaryInnovation]

I will try it right now.  李培宁 ***@***.***  

…

------------------ 原始邮件 ------------------ 发件人: "Marko ***@***.***>; 发送时间: 2023年5月2日(星期二) 晚上6:10 收件人: ***@***.***>; 抄送: ***@***.***>; ***@***.***>; 主题: Re: [njmarko/gnn-polypharmacy-ddi] some questions (Issue #1) Executing split_cv_data.py takes around 20 to 30 minutes. I added requirements.txt to the repo for now. Also, there should be a folds directory in /data/decagon. I added a fix for that now, but you can also create it manually. Let me know if these steps worked for you. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>

[legendaryInnovation]

Executing split_cv_data.py takes around 20 to 30 minutes. I added requirements.txt to the repo for now. Also, there should be a folds directory in /data/decagon. I added a fix for that now, but you can also create it manually.

Let me know if these steps worked for you.

Thank you. The previous ones are very normal. May I ask what the situation is

[njmarko]

Executing split_cv_data.py takes around 20 to 30 minutes. I added requirements.txt to the repo for now. Also, there should be a folds directory in /data/decagon. I added a fix for that now, but you can also create it manually.
Let me know if these steps worked for you.

Thank you. The previous ones are very normal. May I ask what the situation is

Just create a new folder named folds in the decagon folder.
Structure should look something like this
/data/decagon/folds/

[njmarko]

Also, remember to pull new changes from the repo. It should automatically create that folder now. And requirements.txt should be at the top level also. I don't see requirements.txt in your screenshot right now, so you may not have the newest version.

[legendaryInnovation]

Also, remember to pull new changes from the repo. It should automatically create that folder now. And requirements.txt should be at the top level also. I don't see requirements.txt in your screenshot right now, so you may not have the newest version.
Thank you for your reply, but now I'm really sorry that I have a new question.

[njmarko]

Also, remember to pull new changes from the repo. It should automatically create that folder now. And requirements.txt should be at the top level also. I don't see requirements.txt in your screenshot right now, so you may not have the newest version.
Thank you for your reply, but now I'm really sorry that I have a new question.

You should run the train.py with the following parameters, and maybe it will solve your issue
-dataset decagon -input_data_path data/decagon/

[legendaryInnovation]

Thank you very much
I want to know if this error means I need to reinstall Pytorch

AttributeError: module 'torch' has no attribute 'sparse_csc'

[legendaryInnovation]

Thank you very much
I want to know if this error means I need to reinstall Pytorch

AttributeError: module 'torch' has no attribute 'sparse_csc'

[njmarko]

Thank you very much I want to know if this error means I need to reinstall Pytorch

AttributeError: module 'torch' has no attribute 'sparse_csc'

Maybe. Try creating a virtual environment, and then install the requirements that are located in the requirements.txt

[legendaryInnovation]

Thank you very much I want to know if this error means I need to reinstall Pytorch
AttributeError: module 'torch' has no attribute 'sparse_csc'

Maybe. Try creating a virtual environment, and then install the requirements that are located in the requirements.txt

OK，I will try it at once.

[njmarko]

Since you are using PyCharm, you can do like this:

File > Settings > Project: gnn-polypharmacy-ddi > Python Interpreter > Add Interpreter > Add Local Interpreter > Virtualenv Environment

Then simply select a python version. I think I was using python 3.10

[legendaryInnovation]

Since you are using PyCharm, you can do like this:

File > Settings > Project: gnn-polypharmacy-ddi > Python Interpreter > Add Interpreter > Add Local Interpreter > Virtualenv Environment

Then simply select a python version. I think I was using python 3.10

I just reinstalled the environment, but there were more problems. Then, I used the previous environment again and displayed a new problem. May I ask what the problem is? Thank you.
BrokenPipeError: [Errno 32] Broken pipe

[njmarko]

Since you are using PyCharm, you can do like this:
File > Settings > Project: gnn-polypharmacy-ddi > Python Interpreter > Add Interpreter > Add Local Interpreter > Virtualenv Environment
Then simply select a python version. I think I was using python 3.10

I just reinstalled the environment, but there were more problems. Then, I used the previous environment again and displayed a new problem. May I ask what the problem is? Thank you. BrokenPipeError: [Errno 32] Broken pipe

Not sure why that happens. Try to disable logging maybe, by commenting out wandb logging.