IdMotif: An Interactive Motif Identification in Protein Sequences

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Motifs in protein sequences are crucial for understanding protein function and predicting disease.
• Identifying these motifs is essential for biological and medical research.
• Existing methods may lack the interactive visualization needed for effective motif discovery.

Purpose of the Study:

• To present idMotif, a novel visual analytics framework for protein sequence motif identification.
• To enable domain experts to explore, analyze, and visualize motifs in protein sequences.
• To integrate deep learning with interactive visualization for enhanced motif discovery.

Main Methods:

• Developed a visual analytics framework named idMotif.
• Implemented a deep-learning-based method for grouping protein sequences.
• Utilized local explanations of deep-learning model decisions to discover motif candidates.
• Designed interactive linked views for comprehensive analysis.

Main Results:

• The idMotif framework facilitates the exploration and analysis of protein sequence motifs.
• The deep-learning approach effectively groups protein sequences for motif discovery.
• Interactive visualizations aid in understanding motif candidates within protein groups.
• Case studies and expert feedback validate the framework's utility.

Conclusions:

• idMotif enhances the process of protein sequence motif identification for domain experts.
• The integration of deep learning and visual analytics offers a powerful approach to motif discovery.
• This framework aids in predicting diseases and understanding protein functions through sequence analysis.