hips / neural-fingerprint Goto Github PK

Hi, I am wondering that malaria dataset is from real experiments or calculation?

As title says, I've encountered a whole handful of issues attempting to use this software.

After building, I had to manually place packages and libs in my miniconda3 folder. Maybe I could've specified some parameters to install it correctly? Not sure. Here are the real problems:

Attempting to import neuralfingerprint from any IDE or location within terminal outside of neuralfingerprint folder results in: 'util.py' not found. Not a problem right? Just append the path of that folder to my sys.path....

Once fixing the util.py problem, I receive many syntax errors regarding print statements.
E.G.:
print "Total number of weights in the network:", num_weights
^
SyntaxError: Missing parentheses in call to 'print'. Did you mean print("Total number of weights in the network:", num_weights)?

Once fixing those manually, I get the next problem!

File "/mnt/c/Users/came/Documents/neural-fingerprint-master/neural-fingerprint-master/neuralfingerprint/build_convnet.py", line 2, in
from autograd.scipy.misc import logsumexp
ImportError: cannot import name 'logsumexp' from 'autograd.scipy.misc' (/home/azn/miniconda3/lib/python3.9/site-packages/autograd/scipy/misc.py)

I'm no longer interested in putting out fires so I figured I'd post here despite the forum being inactive for the last four years, fingers crossed y'all are still paying attention to this page!

Please let me know if there are any fixes for these issues or if y'all know what's going on here.

I believe it could just be a mismatch between Python3 and Python2, I'm not very thrilled about having to swap python3 to python2 for this so I figured it would be good to ask first.

Many thanks!

Is the metric MAE, MSE or RMSE?

I am so interested in your method that I had to ask another following issue:

Hyperparameter Optimization
"To optimize hyperparameters, we used random search. The hyperparameters
of all methods were optimized using 50 trials for each cross-validation fold. The
following hyperparameters were optimized: log learning rate, log of the initial weight scale, the log
L2 penalty, fingerprint length, fingerprint depth (up to 6), and the size of the hidden layer in the
fully-connected network. Additionally, the size of the hidden feature vector in the convolutional
neural fingerprint networks was optimized."

Could you give me some suggestions about Hyperparameter Optimization?
Or any empirical range about these parameters?

Best regards,
YJ

Hello. I'm trying neuralfingerprint, and have faced strange behaviour:
When i apply the model to csv file, which contains only smiles:
CCC
FFF
i get the result:
CCC,-3.4293943508031028
FCF,-2.6789522776231816

but when i put only CCC, i get another result.
CCC,-3.0120117325667533

If there are same molecules in input file, it gives same results for them, like:
CCC,-3.0120117325667533
CCC,-3.0120117325667533

The predictions are reproductable (dont change after another run), but the exact values depends on the contents of test csv file. I use your example.

smiles = read_smiles(task_params['experiment_data_file'])
result = predict_func(smiles)

I'm wondering if it is a bug or a feature.

Hello,
I was trying to launch script launch_experiments.py in directory experiment_scripts and got error:
Traceback (most recent call last):
File "launch_experiments.py", line 11, in
from nfpexperiment import util
ImportError: No module named nfpexperiment

It seesm that there is not module nfpexperiment. Could you please provide this module?
Thanks in advance.

Hi,

maybe a quite simple question but in your Regression example you pass the

'nll_func': <function neuralfingerprint.util.rmse as the nll_func via build_conv_deep_net to the build_standard_net function. However, I am not sure how the utils.rmse relates to the mean_squared_error function that is being used in the build_standard_net.

My goal is to adapt the example code so that I can do a binary classification.
I tried to replace the default loss from build_standard_net with the binary_cross_entropy. But I think I am missing something because the results do not make sense:

Currently in the file rdkit_utils.py a BitVector is obtained using RDKit, then transformed to a BitString, and then iteratively converted to a np array

AllChem.GetMorganFingerprintAsBitVect(
        m, fp_radius, nBits=fp_length)).ToBitString()

np.array([list(s) for s in A], dtype=int)

Can be written as
DataStructs.ConvertToNumpyArray(AllChem.GetMorganFingerprintAsBitVect(m, fp_radius, Bits=fp_length), np.zeros((1,)))

as seen in
http://www.rdkit.org/Python_Docs/rdkit.DataStructs.cDataStructs-module.html#ConvertToNumpyArray

After the neuralfingerprint install (pip install -e /my_dir/, I'm trying to run the examples (regression.py), but I get a 'utils' not found error...
Can you please help, thanks!

Hello,

I have met a bug about

mol_graph.py", line 79, in graph_from_smiles
raise ValueError("Could not parse SMILES string:", smiles)
ValueError: ('Could not parse SMILES string:', 'CCCCCCCCCCCCN(C)CC(=O)O')"

As described above, the "[N]" in the smiles couldn't be parsed. I wonder if this question is mainly due to the limitation of rdkit package.
Or would you have any better solutions about this bug?

Best regards,
YJ