Welcome to EFHE, an program able to predict if a protein sequence contains EF hands or not. It will also differentiate confirmed and potential EF hands.
The following packages must be installed prior to running EFHE
pip install -U pandas
pip install -U numpy
pip install -U cPickle
pip install -U sklearn
pip install -U tensorflow
pip install -U keras
pip install -U sklearn
Start by cloning the reposititory by using Console or the Github Application. Once installed, make sure the dataset files are accessible.
The EFhandConverter.ipynb converts all the reviewed and unreviewed protein sequences into their seperate EF hands, this program is also responsible for developing a set of false EF hands. The two datasets were combined with a label, 1 representing an EF hand, and 0 representing a non-EF hand. The resulting dataset is named efHandData_neg_pos_interleukins.csv and can be found in the Datasets folder.
Next, using the efHandData_neg_pos_interleukins.csv dataset, the EFhand_CNN+LSTM.ipynb script is run to develop Bi-CNN + LSTM models for predicting EF hands. We manipulated the nb_epoch value to evaluate which model would give the best accuracy. The resulting models are listed in the Results folder.
The final EFHE.ipynb script is used to evaluate protein sequences. Simply input a protein sequence into the sequence value, and ensure the desired epoch model is correctly imported. The result is value containing the number of confirmed ef hands, and the number of potential ef hands. It also includes an prediction value, as well as the sequence and the location it occurs within the protein sequence. A sample using lanmodulin's sequence is provided.