Automated Sub-Cellular Phenotype Classification: An Introduction and Recent Results

Hamilton, N., Pantelic, R., Hanson, K., Fink, J.L., Karunaratne, S. and Teasdale, R.D.

The genomic sequencing revolution has led to rapid growth in sequencing of genes and proteins, and attention is now turning to the function of the encoded proteins. In this respect, microscope imaging of a protein's subcellular location is proving invaluable. High-throughput methods mean that it is now possible to capture images of hundreds of protein localisations quickly and relatively inexpensively, and hence genome-wide protein localisation studies are becoming feasible. However, to a large degree the analysis and localisation classification are still performed by the slow, coarse-grained and possibly biased process of manual inspection. As a step towards dealing with the fast growth in subcellular image data the Automated Sub-cellular Classification system (ASPiC) has been developed: a pipeline for taking cell images, generating statistics and classifying using SVMs. Here, the pipeline is described and correct classification rates of 93.5% and 86.5% on two 8-class subcellular localisation datasets are reported. In addition we present a survey of other important applications of cell image statistics. The complete image sets are being made available with the aim of encouraging further research into automated cell image analysis and classification.

Cite as: Hamilton, N., Pantelic, R., Hanson, K., Fink, J.L., Karunaratne, S. and Teasdale, R.D. (2006). Automated Sub-Cellular Phenotype Classification: An Introduction and Recent Results. In Proc. 2006 Workshop on Intelligent Systems for Bioinformatics (WISB 2006), Hobart, Australia. CRPIT, 73. Boden, M. and Bailey, T. L., Eds. ACS. 67-72.

(from crpit.com) (local if available)