A Graphical Model to Determine the Subcellular Protein Location in Artificial Tissues
Estelle Glory-Afshar1, Elvira Osuna-Highley, Brian Granger
1Center for Bioimage Informatics and Department of Biomedical Engineering, Carnegie Mellon University, Pittsburgh, PA, USA.
Proceedings. IEEE International Symposium on Biomedical Imaging
|September 28, 2011
Summary
Automated methods for protein location proteomics were enhanced using 3D confocal microscopy images of CaCo2 cells. A graphical model approach improved classification accuracy for subcellular protein patterns from 89.2% to 99.6%.
Area of Science:
- Cell Biology
- Proteomics
- Bioimaging
Background:
- Location proteomics systematically analyzes protein subcellular locations.
- Automated methods are crucial for comprehensive analysis of protein location patterns.
- Extending automated analysis to high-resolution tissue images presents challenges.
Purpose of the Study:
- To extend automated subcellular location pattern analysis to high-resolution 3D confocal microscope images of tissues.
- To develop and evaluate a robust classification method for subcellular protein patterns in tissue images.
Main Methods:
- Collected 3D confocal microscope images of polarized CaCo2 cells immunostained for various proteins.
- Developed a three-color staining protocol for proteins of interest, DNA, and actin cytoskeleton.
- Trained a classifier on 11-21 images per protein (9 proteins total) and applied the Prior Updating method.
Main Results:
- A classifier achieved 89.2% accuracy in recognizing subcellular location patterns.
- The Prior Updating method significantly improved classification accuracy to 99.6%.
- Demonstrated the effectiveness of a graphical model approach for pattern classification in tissue images.
Conclusions:
- Automated subcellular location pattern analysis can be successfully extended to high-resolution tissue images.
- Graphical model approaches, specifically Prior Updating, substantially enhance classification accuracy.
- This study validates a powerful method for advancing location proteomics in complex biological samples.
