ProtDCal: A program to compute general-purpose-numerical descriptors for sequences and 3D-structures of proteins

Area of Science:

• Bioinformatics
• Computational Biology
• Structural Biology

Background:

• Exponential growth in protein databases necessitates advanced methods for sequence-structure-function analysis.
• Computational tools are crucial for mining patterns and learning models relating protein structure and function.
• Efficiently extracting numerical descriptors for protein sequence and structure remains a challenge.

Purpose of the Study:

• Introduce ProtDCal, a novel computational software suite for generating comprehensive protein descriptors.
• To demonstrate the added value of ProtDCal's sequence-based and 3D-structural features.
• To showcase the utility of ProtDCal features in predicting biological properties like N-linked glycosylation sites.

Main Methods:

• Development of ProtDCal software with a Java-based graphical user interface.
• Generation of tens of thousands of sequence-based and 3D-structural descriptors.
• Utilizing Principle Component Analysis and Shannon entropy tests for descriptor evaluation.
• Training and evaluating prediction models for N-linked glycosylation sites.

Main Results:

• ProtDCal's sequence descriptors offer novel information not present in existing servers.
• 3D-structure-based features provide complementary and non-redundant information, enhancing overall protein encoding.
• Prediction models using ProtDCal features achieved favorable classification performance for N-linked glycosylation sites.

Conclusions:

• ProtDCal is a user-friendly, cross-platform tool for general-purpose protein encoding.
• The software enhances feature diversity through local and group-based encoding.
• ProtDCal facilitates applications in protein classification, similarity analysis, and function prediction.