Loading...

Drug-target Interaction Prediction through Learning Methods for SARS-COV2 Based on Sequence and Structural Data

Gheysari, Maryam | 2022

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 57086 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Koohi, Somayyeh
  7. Abstract:
  8. Predicting the binding affinity of drug molecules and proteins is one of the most important stages of drug discovery, development and screening, for which numerous laboratory, simulation and computational solutions have been provided. Laboratory and simulation methods require molecular structures, are time and financial expense, and computational methods do not provide accurate predictions. Therefore, the use of deep neural networks in extracting features from data with a simpler and more accessible structure of protein sequences and molecules solves these challenges with lower cost and higher accuracy. In this article, the use of a new molecular sequence named selfies, which has solved the defects of the previous structures, including complete mismatch with valid molecules and unfavorable modeling of the local useful features of the side chain structure and molecular rings, in the problem of predicting the binding affinity of drug and protein, it has been suggested. This molecular structure along with the extracted protein sequence feature vectors from the ELMo model has been used in a model based on deep convolutional neural networks. Applying this molecular structure to common data sets, as well as using the transfer learning technique to extract features from the pre-trained model, more accuracy compared to the previous complex models, in the mean square error evaluation criterion and the concordance index of competitive results in a neural network architecture
  9. Keywords:
  10. Neural Network ; Convolutional Neural Network ; Binding Affinity ; Transfer Learning ; Pharmaceutical Compound ; Protein Sequence ; Embedding from Language Model (ELMo) ; Selfies Model ; Drug-Target Interaction ; Feature Vector

 Digital Object List

 Bookmark

No TOC