Protein secondary structure assignments using ML

Researchers seek to understand computer-generated protein models by identifying their structural components. Categorizing amino acid residues as Helix, Strand, or Coil types is the process used for this identification. This categorization can be challenging if certain atoms are missing or only alpha carbon locations are available. To overcome this challenge, researchers have developed various techniques over the past few decades, with machine learning being the latest approach. Recently, a classifier that utilizes neural networks to identify secondary structure types of proteins was introduced. The classifier was exclusively trained using Cα coordinates and the Keras (TensorFlow) library, and neural network input features were extracted from raw coordinates using the BioShell toolkit. By carefully selecting input features, the method achieved an accuracy of over 97%, surpassing existing methods. The research has significant implications for understanding the structure and function of proteins in biological processes, including drug discovery and disease treatment. Moreover, the success of this approach demonstrates the potential of machine learning techniques in protein structure determination, paving the way for future advancements in this field. Further research can explore the application of this approach to other types of proteins and the integration of additional information sources to improve accuracy even further.