DL_STRUCT

This is a tutorial on how to use geometric deep learning on protein structures.

It's organized in four parts:

PART 1:

• A brief introduction on Deep Neural Networks (DNN)
• A brief introduction on Convolutional Neural Networks (CNN): Convolution, Convolutional layers, Pooling layers, Batch normalization, CNN Architectures, ...
• Properties of CNN: such as translational-invariant
• Machine learning concepts: such as Overfitting, Dropouts, and ...
• An application: classification of handwritten digits
• Training and evaluating a model
  Google Colab:
  https://colab.research.google.com/drive/1BAc6Sh1lhy06_B56JBRCVY-JWlk743qO?usp=sharing

PART 2:

• Analyzing amino acids and their environments
• Translation and rotation-invariant representation of the local environment
• An application: prediction of the amino acid with and without considering their surroundings
• Training and evaluating a model
• A discussion on the results
  Google Colab:
  https://colab.research.google.com/drive/1cQ3POBsdfUM1gg2VN4hX6Ps8ZIC1B7n_?usp=sharing

PART 3:

• An introduction on protein-protein interaction (PPI) and the concept of binding affinity
• An introduction on point mutations
• Effects of point mutations on PPI
• Analysis of the wild-type and mutant protein structures
• An application: prediction of changes in PPI binding affinity upon mutation using rotation and translation-invariant representation
• Training and evaluating a model
• Analysis of the results
  Google Colab:
  https://colab.research.google.com/drive/1Isfgq9F_9rzgaIT6n4R7uFDkHGqGvfA2?usp=sharing

PART 4 (Coming soon):

• A brief introduction on Graph Convolutional Networks (GCN)
• A brief introduction on Graph Attention Networks (GAT)
• An application