Using published pathway figures in enrichment analysis and machine learning.

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Pathway databases are crucial for understanding biological processes and disease mechanisms.
• Existing databases like Gene Ontology offer breadth but may lack detailed mechanistic diagrams and direct literature links.
• There is a need for pathway resources that integrate visual, mechanistic information with extensive literature support.

Purpose of the Study:

• To introduce Pathway Figure OCR (PFOCR) as a novel pathway database.
• To assess PFOCR's utility in disease research compared to established pathway databases.
• To demonstrate PFOCR's unique advantages in advanced analytical applications, including cancer subtype and grade prediction.

Main Methods:

• Developed PFOCR, a pathway database incorporating OCR technology for extracting information from pathway diagrams.
• Assessed disease coverage of PFOCR against other popular pathway databases.
• Evaluated analytical applications of PFOCR, including common pathway analyses and advanced case studies.

Main Results:

• PFOCR approaches the breadth and depth of Gene Ontology.
• PFOCR provides rich, mechanistic diagrams and direct literature support.
• PFOCR demonstrated unique advantages in cancer subtype and grade prediction analyses.

Conclusions:

• PFOCR is a valuable new resource for disease research, offering a unique combination of detailed pathway diagrams and literature support.
• PFOCR's capabilities extend beyond traditional pathway analysis, showing promise in complex predictive modeling for diseases like cancer.