Error in sequence analysis generated in source code about lstm_peptides HOT 5 CLOSED

hi zhouhao-learning
first of all: great that you found our code and a use case for it.
I'm sorry for the inconvenience. an external module, modlamp, has changed in the meantime. I have fixed the bug there. yet, you will need to do the following to make it work:

1. pip install --update modlamp
2. update the LSTM_peptides repository (in the LSTM_peptides folder type "git pull origin master")

let me know if you succeed

from lstm_peptides.

@alexarnimueller
Hello,
First of all, thank you for your reply, but I cloned your code, updated modlamp, and used Python 2.7. When I executed the command python LSTM_peptides.py--epochs 2, I got another error:

Traceback (most recent call last):
  File "LSTM_peptides.py", line 793, in <module>
    _save_flags("./" + FLAGS.name + "/flags.txt")
  File "LSTM_peptides.py", line 78, in _save_flags
    f.write(k + ": " + str(v.value) + "\n")
AttributeError: 'bool' object has no attribute 'value'

I resolved this error by modifying your LSTM_peptides.py file. Here is the correct code after the modification.

def _save_flags(flgs, filename):
    with open(filename, 'w') as f:
        f.write("Used flags:\n-----------\n")
        for k, v in flgs.__dict__['__flags'].items():
            f.write(k + ": " + str(v) + "\n")

Finally, I still can't run in Python 3 environment. I think you should change some code so that it can be executed in Python 3 environment, because Python 2 is not updated now. I hope you can take my advice. Thank you.

from lstm_peptides.

The code runs very well on python2.7 and I will keep it like that. Why don't you make an environment for it?
Concerning the flags: newer tensorflow versions don't support your change. So better keep my way of saving the flags and just update your tensorflow

from lstm_peptides.

Well, I just recommend that you change it to Python version 3, because I believe many people will prefer to use Python 3 if they see your code later, because Python 3 is more active and python 2 is no longer updated, and the tensofrlow 1.3 you provide does not make your code work properly, it will be at the time of model creation. error

from lstm_peptides.

I don't want to argue any longer about it. If there is a bug, let me know where and I'll fix it. The code runs without error for me

from lstm_peptides.