From homogeneous to heterogeneous network alignment via colored graphlets.

Area of Science:

• Computational Biology
• Network Science
• Bioinformatics

Background:

• Network alignment (NA) is crucial for comparing biological networks and identifying conserved regions.
• Existing NA methods are limited to homogeneous networks (single node/edge types).
• The rise of heterogeneous networks necessitates new alignment approaches.

Purpose of the Study:

• To extend state-of-the-art homogeneous NA methods (WAVE, MAGNA++, SANA) to handle heterogeneous networks.
• To develop novel measures for heterogeneous node similarity and edge conservation.
• To evaluate the performance of the proposed heterogeneous NA methods.

Main Methods:

• Extended homogeneous graphlets to heterogeneous graphlets for node similarity calculation.
• Developed a new heterogeneous node similarity measure.
• Adapted the S³ metric for heterogeneous edge conservation.
• Applied these novel measures within existing NA frameworks (WAVE, MAGNA++, SANA).

Main Results:

• Proposed heterogeneous NA methods achieve higher alignment quality compared to their homogeneous counterparts.
• The new methods demonstrate improved robustness against noise in biological network data.
• Evaluations on synthetic and real-world biological networks validate the approach.

Conclusions:

• The developed heterogeneous NA methods represent a significant advancement for analyzing complex biological networks.
• These methods provide more accurate and reliable insights into conserved network structures.
• The approach offers a robust solution for network alignment in the presence of data heterogeneity and noise.