Feature Representation and Learning Methods with Applications in Protein Secondary Structure

For many years, protein structures have been determined using time-consuming, high-cost wet experiments. Given that the structure of the protein is derived from the protein sequence, many new computational methods have been proposed to effectively solve the problem of protein structure prediction. The rise of machine learning methods, especially deep learning, to predict protein secondary structures in recent years has greatly promoted the development of this field. Such methods can not only make better use of exponentially growing massive protein sequence data , but are also able to automatically mine complex and latent patterns hidden in the data, thereby continuously improving the prediction accuracy of protein secondary structures. Although significant progress has been made, we still face challenges in how to predict protein secondary structures directly from protein sequences with improved accuracy.

The latest advancements in numerical representation of protein sequence feature representation, combined with statistical and machine learning methods, are expected to discover or summarize the sequence characteristics that determine the secondary structure of a protein. These developments are also expected to improve universal high-precision prediction methods applicable to protein secondary structure, to help better understand the sequence model characteristics of some special protein secondary structures (such as transmembrane protein), and to use more accurate protein secondary structure predictors for protein function, protein folding, or drug discovery, etc.

This Research Topic welcomes Original Research Articles and Reviews on the topics of:

• Databases or new benchmark datasets relative to protein secondary structures
• New machine learning methods or tools for protein secondary structures prediction
• New sequence features representation methods for protein secondary structures prediction
• Prediction of the specific secondary structures, such as transmembrane proteins, π helices and 310 helices, etc.
• Application of protein secondary structures prediction in protein function, protein-protein interaction, protein-RNA interaction, drug discovery and other fields.
• Using protein modelling to uncover and explore structure-function relationships and identify new, non-natural protein function.