Learning the rules of collective cell migration using deep attention networks.

Area of Science:

• Biophysics
• Cell Biology
• Computational Biology

Background:

• Multicellular organisms rely on coordinated cell movement.
• Understanding cell-cell interaction dynamics in collective migration is challenging.
• Existing methods struggle to intuitively capture high-dimensional interaction data.

Purpose of the Study:

• To develop a method for extracting collective cell migration rules.
• To analyze and compare migration behaviors in different tissue types.
• To visualize and interpret cell-cell interaction dynamics.

Main Methods:

• Application of deep attention networks to cell migration trajectory data.
• Analysis of epithelial, endothelial, and metastatic breast cancer cell behaviors.
• Generation of attention maps to visualize neighbor influence on cell decisions.

Main Results:

• Distinct attention patterns identified for each tissue type.
• Endothelial cells show focused attention on forward neighbors.
• Epithelial tissues exhibit broader forward sector influence.
• Mesenchymal cells display symmetric, uncoordinated attention patterns.
• Attention networks adapt to biophysical context like crowding.

Conclusions:

• Deep attention networks can decipher complex collective cell migration rules.
• These networks provide intuitive biological insights without prior modeling assumptions.
• The approach offers a powerful tool for studying cellular systems.