https://github.com/Bjoux2/DeepDTIs_DBN

Deep learning-based drug-target interaction prediction / Deep belief net (DBN) based on Theano
https://github.com/Bjoux2/DeepDTIs_DBN

Last synced: 3 months ago
JSON representation

Deep learning-based drug-target interaction prediction / Deep belief net (DBN) based on Theano

• Host: GitHub
• URL: https://github.com/Bjoux2/DeepDTIs_DBN
• Owner: Bjoux2
• License: gpl-3.0
• Created: 2016-04-27T06:36:57.000Z (over 8 years ago)
• Default Branch: master
• Last Pushed: 2019-02-26T04:25:59.000Z (over 5 years ago)
• Last Synced: 2024-07-16T23:02:38.705Z (4 months ago)
• Language: Python
• Homepage:
• Size: 11.8 MB
• Stars: 47
• Watchers: 8
• Forks: 21
• Open Issues: 4
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-drug-discovery - [Python Reference

README

        
# DeepDTIs

Please cite 'Deep learning-based drug-target interaction prediction'.

The Deep belief net (DBN) code was rewritten from www.deeplearning.net 

The code in 'code_sklearn-like' is recommended, the usage of the DBN here is similar to sklean:

## Pseudo-example:

from DBN_wm import DBN

dbn_classifier = DBN()

dbn_classifier.pretraining(train_x) 

dbn_classifier.finetuning(train_x, train_y, valid_x, valid_y)    # the valid set is used to optimize the parameters

y_pred = dbn_classifier.predict(test_y)

>>>More detaild example, see test_DBN.py

>>>Please note that: the calculated data is very large (>4GB), we could not upload the calculated data. If you need the data, please follow the Data section (download molecules and proteins from Drugbank & calculate features using Biotriangle web platform or other python packages) in the paper to construct the training data.

## Dependencies:

1), Python 2.7, latest version

2), Theano, latest version

## Further reading: 

1), Deep-Learning-in-Bioinformatics-Papers-Reading-Roadmap

https://github.com/Bjoux2/Deep-Learning-in-Bioinformatics-Papers-Reading-Roadmap