MNNN (Multi-scale Neighborhood-based Neural Network) is an example code for predicting dihedral angles using only primary protein sequences. It is very useful for computing the 3D structural information of known and unknown protein sequence.
python>=3.6.
tensorflow==1.8.
sklearn==0.19.1
matplotlib==2.0.2
python MNNN/food_train.py
(1) paste target data into MNNN/foodData/testData.txt. (format: "ABC... %name", as example)
(2) run MNNN/foodData/interGen.py to reformat the input data.
(3) run MNNN/inferrence.py to get model output.
(4) run MNNN/generateOutput.py to reformat output data based on configurations.
(5) run MNNN/results/npyRes/printPreds.py to display the output model predictions, output path: MNNN/results/npyRes/output.txt