Scholarlink

Drug discovery and biological computation depend on the ability to forecast improvements in molecular affinity prediction. An inventive strategy was created to enhance molecule binding affinity predictions by fusing cutting-edge computational techniques with a hybrid model that made use of deep learning. With the use of this inventive technique, protein sequences characterized by amino acid chains were connected to drug structure representations in SMILES format. The method interpreted complicated patterns in chemical and biological systems by using a convolution-bound BiLSTM model and a comprehensive bioinformatics approach. Through a confidence interval (CI) of 0.86, it accurately predicts drug-protein interactions and is useful for capturing crucial structural and relational properties. These results provided a solid framework for understanding molecular interactions and illustrated the model's overall effectiveness in comparison to earlier computational aided molecular biology methods.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multimodal Deep Learning for Enhanced Prediction of Molecular Binding Affinities Integrating Chemical Structures and Protein Sequences

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • L. Prasika,
  • G. R. Karish Prajaishma,
  • M. Yoga Vardhani

摘要

Drug discovery and biological computation depend on the ability to forecast improvements in molecular affinity prediction. An inventive strategy was created to enhance molecule binding affinity predictions by fusing cutting-edge computational techniques with a hybrid model that made use of deep learning. With the use of this inventive technique, protein sequences characterized by amino acid chains were connected to drug structure representations in SMILES format. The method interpreted complicated patterns in chemical and biological systems by using a convolution-bound BiLSTM model and a comprehensive bioinformatics approach. Through a confidence interval (CI) of 0.86, it accurately predicts drug-protein interactions and is useful for capturing crucial structural and relational properties. These results provided a solid framework for understanding molecular interactions and illustrated the model's overall effectiveness in comparison to earlier computational aided molecular biology methods.