Loading...

Hydrophobocity Prediction of Molecules Using Graph Neural Networks

Feili, Mohammad | 2023

2 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 56982 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Shamloo, Amir
  7. Abstract:
  8. Prediction of the hydrophobicity of molecules using graph neural networks is one of the advanced methods for analyzing the physicochemical properties of molecules. In this method, the molecular graph structure and its features are used as inputs to neural networks to provide predictions about the hydrophobicity of molecules. This method is currently applied in the pharmaceutical industry, green chemistry, and other research fields. By using this approach, molecules with strong hydrophobic properties can be identified and used in the development of new drugs. Generally, graph neural networks are highly suitable for predicting the properties of molecules and other various applications in the world of chemistry and pharmaceuticals. In this article, the hydrophobicity prediction of molecules using graph neural networks has been investigated. Specifically, convolutional graph layers and graph attention layers have been utilized. This approach, taking into account the hydrophobic (lipophilic) nature of a molecule that can encompass a part of a drug, is used so that the information from neighboring nodes for each node uses attention to some neighbors over others. This new and precise approach in predicting the physical properties of molecules leads to its application in the chemical and pharmaceutical industries. In contrast to the deep learning methods based on Euclidean space that have been used to date, graph neural networks perform well on data that has a graphical structure. This type of deep learning model performs well in extracting important information from data with a graphical structure
  9. Keywords:
  10. Molecular Properties Prediction ; Hydrophobicity ; Deep Learning ; Graph Neural Network ; Convolutional Neural Network

 Digital Object List