Automated Proteome-Wide Determination of Subcellular Location Using High Throughput Microscopy

Area of Science:

• Cell Biology
• Computational Biology
• Microscopy

Background:

• Identifying subcellular locations of proteins is crucial for understanding cellular functions.
• Fluorescence microscopy of tagged proteins in living cells is a primary method for this identification.
• Automated image analysis systems have been developed to interpret these microscopy images.

Purpose of the Study:

• To demonstrate the application of automated image interpretation methods to large-scale microscopy datasets.
• To develop methods for discovering and representing patterns of subcellular protein localization.
• To integrate these findings into systems biology efforts.

Main Methods:

• Automated interpretation of fluorescence microscopy images of tagged proteins (e.g., GFP-tagged yeast, immunocytochemistry in human tissues).
• Clustering algorithms to group proteins with similar subcellular localization patterns.
• Learning generative models to represent and communicate identified subcellular location patterns.

Main Results:

• Automated systems perform comparably to or better than visual inspection for subcellular location identification.
• Methods are applicable to diverse large-scale image collections from yeast and human tissues.
• Successful clustering of mouse proteins into distinct subcellular location families based on shared patterns.

Conclusions:

• Automated image analysis of tagged proteins provides a powerful approach for proteome-wide subcellular localization.
• Generative models can effectively represent and communicate learned subcellular location patterns.
• Integration of high-throughput microscopy and automated model building enhances systems biology research.