Skip to content
No description, website, or topics provided.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
Evaluation Files
Hermes Results
Structure Layer Neural Network Weights - Models
1Letter.txt
Hermes_Main.py
Houston.py
LICENSE
README.md
Sample_DNA_Seqs_For_Hermes.fa
triplet.txt

README.md

Hermes Preprint: https://www.biorxiv.org/content/10.1101/640656v1

~ Due to GitHub file size limits, this repository contains a version of Hermes without the mechanism to deal with individual Protein layer server crashes. The full version is avalible as a zipped folder from Google Drive: https://drive.google.com/open?id=1sy_DXgULBPtS4TAYqY2fyF9TnbqkOpDj

  1. Download the master branch
  2. Run the file "Houston.py"
  3. Results are given as output to the console

Dependencies

Python Version: 3.6.2

Required Python Packages

Name: Tensorflow Version: 1.5.0 URL Access: https://pypi.org/project/tensorflow/1.5.0/

Name: beautifulsoup4 Version: 4.6.0 URL Access: https://pypi.org/project/beautifulsoup4/4.6.0/

Name: requests Version: 2.17.3 URL Access: https://pypi.org/project/requests/2.17.3/

Name: imaplib2 Version: 2.45.0 URL Access: https://pypi.org/project/imaplib2/2.45.0/

Name: numpy Version: 1.13.1 URL Access: https://pypi.org/project/numpy/1.13.1/

Name: Keras Version: 2.1.3 URL Access: https://pypi.org/project/Keras/2.1.3/

Files Provided

  1. Houston.py • The laucher file

  2. Hermes_Main.py • The core pipeline, intitated by the launcher

  3. Sample_DNA_Seqs_For_Hermes.Fa • Multiple DNA sequences, given in FASTA format, to provide an initial run

  4. triplet.txt • The codon table as a text file, which is used internally

  5. 1Letter.txt • The one-letter amino acid codes as a text file, again used internally

  6. /Structure Layer Neural Network Weights - Models/Structure_Layer_BiLSTM.tfl • The Bidirectional LSTM (BiLSTM) classifier weights & biases of the Structure layer

  7. /Structure Layer Neural Network Weights - Models/Structure_Layer_CNN.tfl • The Convolutional Neural Network (CNN) classifier weights & biases of the Structure layer

  8. /Structure Layer Neural Network Weights - Models/Structure_Layer_GRU.tfl • The Gated Recurrent Unit (GRU) classifier weights & biases of the Structure layer

  9. /Structure Layer Neural Network Weights - Models/Structure_Layer_MLP.pickle • The Multilayer Perceptron (MLP) classifier weights & biases of the Structure layer

  10. /Structure Layer Neural Network Weights - Models/Structure_Layer_RNN.tfl • The Recurrent Neural Network (RNN) classifier weights & biases of the Structure layer

  11. /Structure Layer Neural Network Weights - Models/Hermes_Aggregator_Model.tfl • The final aggregator network weights & biases, a BiLSTM classifier

  12. /Hermes Results/Hermes_Result_d23481bb7250fb.txt • A text file, serving as a single example within the cashe system, that is used to save protein layer predictions

  13. /Hermes Results/Hermes_Result_facff3fc114cb2.txt • Another text file, serving as a second example within the cashe system, that is used to save protein layer predictions

  14. /Evaluation Files/JPRED Blind Test/All JPRED Blind Test Predictions.txt • A text file of all JPRED Blind test proteins, with all predictions, respective Q3's, and the experimentally determined DSSP-classified structure

  15. /Evaluation Files/JPRED Blind Test/All JPRED Blind Test NEFF & Identity Values • A text file of all NEFF & identity values for all the JPRED Blind test proteins

  16. /Evaluation Files/CASP11/All CASP11 Predictions.txt • A text file of all CASP11 proteins, with all predictions, respective Q3's, and the experimentally determined DSSP-classified structure

  17. /Evaluation Files/CASP11/All CASP11 NEFF & Identity Values • A text file of all NEFF & identity values for all the CASP11 proteins

  18. /Evaluation Files/7-Fold Cross Validation/7-Fold CV Model 1.tfl-7-Fold CV Model 7.tfl • Seven files, each a set of weights & biases for the respective cross-validation model

You can’t perform that action at this time.