Hello,

I tried to train the model but I got this error

Traceback (most recent call last): File "/opt/anaconda3/bin/molencoder", line 9, in load_entry_point('molencoder==0.1a0', 'console_scripts', 'molencoder')() File "/opt/anaconda3/lib/python3.5/site-packages/molencoder-0.1a0-py3.5.egg/molencoder/cli/main.py", line 32, in main File "/opt/anaconda3/lib/python3.5/site-packages/molencoder-0.1a0-py3.5.egg/molencoder/cli/main.py", line 37, in args_func File "/opt/anaconda3/lib/python3.5/site-packages/molencoder-0.1a0-py3.5.egg/molencoder/cli/parser_train.py", line 31, in func File "/opt/anaconda3/lib/python3.5/site-packages/torch/nn/modules/module.py", line 136, in apply module.apply(fn) File "/opt/anaconda3/lib/python3.5/site-packages/torch/nn/modules/module.py", line 137, in apply fn(self) File "/opt/anaconda3/lib/python3.5/site-packages/molencoder-0.1a0-py3.5.egg/molencoder/utils.py", line 180, in initialize_weights File "/opt/anaconda3/lib/python3.5/site-packages/torch/nn/modules/module.py", line 238, in getattr type(self).name, name)) AttributeError: 'GRU' object has no attribute 'weight'

I had to install pytorch via the conda install. The version is pytorch-0.1.12 and I am using python 3.5.2.

Thanks

(dl3) dendisuhubdy@grok-machine:~/data$ molencoder train --dataset molecules/chembl22.h5 --cuda
Epoch 0:
t = 100, loss = 0.0532
t = 200, loss = 0.0467
t = 300, loss = 0.0473
t = 400, loss = 0.0473
t = 500, loss = 0.0468
t = 600, loss = 0.0469
t = 700, loss = 0.0467
t = 800, loss = 0.0474
t = 900, loss = 0.0470
t = 1000, loss = 0.0456
t = 1100, loss = 0.0481
t = 1200, loss = 0.0443
t = 1300, loss = 0.0485
t = 1400, loss = 0.0474
t = 1500, loss = 0.0465
t = 1600, loss = 0.0469
t = 1700, loss = 0.0460
t = 1800, loss = 0.0453
t = 1900, loss = 0.0476
t = 2000, loss = 0.0466
t = 2100, loss = 0.0454
t = 2200, loss = 0.0471
t = 2300, loss = 0.0462
t = 2400, loss = 0.0458
t = 2500, loss = 0.0464
t = 2600, loss = 0.0475
t = 2700, loss = 0.0474
t = 2800, loss = 0.0442
t = 2900, loss = 0.0463
t = 3000, loss = 0.0455
t = 3100, loss = 0.0458
t = 3200, loss = 0.0457
t = 3300, loss = 0.0470
t = 3400, loss = 0.0467
t = 3500, loss = 0.0450
t = 3600, loss = 0.0473
t = 3700, loss = 0.0471
t = 3800, loss = 0.0471
t = 3900, loss = 0.0461
t = 4000, loss = 0.0497
An unexpected error has occurred with osprey (version 0.1a), please
consider sending the following traceback to the osprey GitHub issue tracker at:
        https://github.com/cxhernandez/molencoder/issues

Traceback (most recent call last):
  File "/u/suhubdyd/.conda/envs/dl3/bin/molencoder", line 11, in <module>
    load_entry_point('molencoder==0.1a0', 'console_scripts', 'molencoder')()
  File "/u/suhubdyd/.conda/envs/dl3/lib/python3.6/site-packages/molencoder-0.1a0-py3.6.egg/molencoder/cli/main.py", line 33, in main
  File "/u/suhubdyd/.conda/envs/dl3/lib/python3.6/site-packages/molencoder-0.1a0-py3.6.egg/molencoder/cli/main.py", line 38, in args_func
  File "/u/suhubdyd/.conda/envs/dl3/lib/python3.6/site-packages/molencoder-0.1a0-py3.6.egg/molencoder/cli/parser_train.py", line 55, in func
  File "/u/suhubdyd/.conda/envs/dl3/lib/python3.6/site-packages/molencoder-0.1a0-py3.6.egg/molencoder/utils.py", line 89, in validate_model
ValueError: too many values to unpack (expected 2)

Hi,

Thanks for providing this great code.

I am trying to use my own dataset of SMILES strings, I couldn't find the script for preprocessing these raw strings to the input features (I notice the data you provided is already being preprocessed). If you have any, Could you point that out to me? Thanks!

Hi,

I just ran into a problem installing on anaconda python3.6 on ubuntu, after python setup.py install I got the following error:

Installed /home/scott/anaconda3/lib/python3.6/site-packages/molencoder-0.1a0-py3.6.egg
Processing dependencies for molencoder==0.1a0
Searching for yaml
Reading https://pypi.python.org/simple/yaml/
No local packages or working download links found for yaml
error: Could not find suitable distribution for Requirement.parse('yaml')

Though changing yaml to pyyaml in the requirements.txt file seems to have fixed it.

Not sure if this is due to my python installation or a typo, but thought it might be useful to know.

This PR includes miscellaneous changes to the code base. Of note:

• [x] Replace ReLUs with SELU for improved performance
• [x] Add Xavier initialization
• [x] Add reference model which has ~98% on validation set

Hi,

when executing

molencoder download --dataset  zinc12

I am getting a 404 error

  File "/home/sanchezg/oxford/tools/molencoder/molencoder/cli/main.py", line 37, in args_func
    args.func(args, p)
  File "/home/sanchezg/oxford/tools/molencoder/molencoder/cli/parser_download.py", line 62, in func
    urllib.request.urlretrieve(uri, fd.name, reporthook=update)
  File "/home/sanchezg/app/anaconda3/envs/deepMols/lib/python3.7/urllib/request.py", line 247, in urlretrieve
    with contextlib.closing(urlopen(url, data)) as fp:
  File "/home/sanchezg/app/anaconda3/envs/deepMols/lib/python3.7/urllib/request.py", line 222, in urlopen
    return opener.open(url, data, timeout)
  File "/home/sanchezg/app/anaconda3/envs/deepMols/lib/python3.7/urllib/request.py", line 531, in open
    response = meth(req, response)
  File "/home/sanchezg/app/anaconda3/envs/deepMols/lib/python3.7/urllib/request.py", line 641, in http_response
    'http', request, response, code, msg, hdrs)
  File "/home/sanchezg/app/anaconda3/envs/deepMols/lib/python3.7/urllib/request.py", line 569, in error
    return self._call_chain(*args)
  File "/home/sanchezg/app/anaconda3/envs/deepMols/lib/python3.7/urllib/request.py", line 503, in _call_chain
    result = func(*args)
  File "/home/sanchezg/app/anaconda3/envs/deepMols/lib/python3.7/urllib/request.py", line 649, in http_error_default
    raise HTTPError(req.full_url, code, msg, hdrs, fp)
urllib.error.HTTPError: HTTP Error 404: NOT FOUND

Hi! Thanks for the implementation

I think there might be some problem when implementing the conv1d layer here:

https://github.com/cxhernandez/molencoder/blob/1d7e208713d8e97683650b1bbc37a9fa298b4ce0/molencoder/models.py#L56

I think the definition of o' andi' are swapped. The input filter dimension should the number of possible characters (35), and `c' should be the padded SMILES sequence length.

Just want to confirm this detail. I can submit a PR if that is correct.

RuntimeError: Error(s) in loading state_dict for MolEncoder: size mismatch for dense_1.0.weight: copying a param with shape torch.Size([435, 290]) from checkpoint, the shape in current model is torch.Size([435, 270]).

Confirmed by another user in the comments of this article, who suspects the charset might be different? https://iwatobipen.wordpress.com/2018/02/18/mol-encoder-with-pytorch/#comments

I was trying to download the chembl22 dataset and preprocess but got the following error.

$ molencoder download --dataset chembl22
molencoder/.venv/lib/python3.7/site-packages/pandas-1.0.3-py3.7-linux-x86_64.egg/pandas/compat/__init__.py:117: UserWarning: Could not import the lzma module. Yo
ur installed Python is incomplete. Attempting to use lzma compression will result in a RuntimeError.
  warnings.warn(msg)
Downloading Dataset...
 98% |###########################################################################################################################################  | ETA:   0:00:00  18.0 MiB/s
Loading Dataset...
Processing Dataset...
Saving Dataset...
Error: generator raised StopIteration
100% |#############################################################################################################################################| Time:  0:11:34 193.1 KiB/s

Any idea how to fix it?

When I try to run the code, the following error was observed. Could you help with fixing the error? Many thanks. It relates to the absence of a test_data group in the dataset. Could possibly have something to do with the updated database.