Limits...
CYCLoPs: A Comprehensive Database Constructed from Automated Analysis of Protein Abundance and Subcellular Localization Patterns in Saccharomyces cerevisiae.

Koh JL, Chong YT, Friesen H, Moses A, Boone C, Andrews BJ, Moffat J - G3 (Bethesda) (2015)

Bottom Line: Changes in protein subcellular localization and abundance are central to biological regulation in eukaryotic cells.Quantitative measures of protein dynamics in vivo are therefore highly useful for elucidating specific regulatory pathways.The images depict the localization and abundance dynamics of more than 4000 proteins under two chemical treatments and in a selected mutant background.

View Article: PubMed Central - PubMed

Affiliation: The Donnelly Centre, University of Toronto, Toronto, Ontario, Canada, M5S3E1.

No MeSH data available.


Related in: MedlinePlus

Diagram illustrating the ensemble of 60 binary classifiers for protein localization and quantification (modified from Chong et al. 2015). Only cell images that were not filtered by the quality-control classifiers for dead cells and “ghost” objects were further classified. All cells were first classified into different cell-cycle stages using the unbudded and budded classifiers. The rest of the ensemble is organized into 20 subgroups. For some classifier groups, e.g., Cortical Patches and Cell Periphery, budded and unbudded cells were separately tested. The results from each subgroup of binary classifiers e.g., CoP, CoP-Mito, and CoP-Cyto, were consolidated with Bagging. The circles denote the percentage of 1,057,871 cells in the wild-type WT1 experiment that were assigned to each localization class, with darker green indicating a greater percentage.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4478550&req=5

fig2: Diagram illustrating the ensemble of 60 binary classifiers for protein localization and quantification (modified from Chong et al. 2015). Only cell images that were not filtered by the quality-control classifiers for dead cells and “ghost” objects were further classified. All cells were first classified into different cell-cycle stages using the unbudded and budded classifiers. The rest of the ensemble is organized into 20 subgroups. For some classifier groups, e.g., Cortical Patches and Cell Periphery, budded and unbudded cells were separately tested. The results from each subgroup of binary classifiers e.g., CoP, CoP-Mito, and CoP-Cyto, were consolidated with Bagging. The circles denote the percentage of 1,057,871 cells in the wild-type WT1 experiment that were assigned to each localization class, with darker green indicating a greater percentage.

Mentions: The ensLOC framework comprises several steps (Figure 1). We first segmented the cells from micrographs obtained from our wild-type screen. A total of 430 image features, including area, shape, intensity, texture, and Zernike moments (projections of image functions based on a set of orthogonal Zernike polynomials; Teague 1979) were extracted from the segmented cells. For each classifier, we used the χ2 test of independence (Liu and Setiono 1995) to select features that best discriminated the positive from the negative training instances. The filtered features were then used as input to construct the linear Support Vector Machine classifier (Platt 1998). Seventy thousand instances of cell images representative of the morphological signatures of 16 subcellular localizations were handpicked. The distribution of these training instances is shown in Figure 2 and Table 2.


CYCLoPs: A Comprehensive Database Constructed from Automated Analysis of Protein Abundance and Subcellular Localization Patterns in Saccharomyces cerevisiae.

Koh JL, Chong YT, Friesen H, Moses A, Boone C, Andrews BJ, Moffat J - G3 (Bethesda) (2015)

Diagram illustrating the ensemble of 60 binary classifiers for protein localization and quantification (modified from Chong et al. 2015). Only cell images that were not filtered by the quality-control classifiers for dead cells and “ghost” objects were further classified. All cells were first classified into different cell-cycle stages using the unbudded and budded classifiers. The rest of the ensemble is organized into 20 subgroups. For some classifier groups, e.g., Cortical Patches and Cell Periphery, budded and unbudded cells were separately tested. The results from each subgroup of binary classifiers e.g., CoP, CoP-Mito, and CoP-Cyto, were consolidated with Bagging. The circles denote the percentage of 1,057,871 cells in the wild-type WT1 experiment that were assigned to each localization class, with darker green indicating a greater percentage.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4478550&req=5

fig2: Diagram illustrating the ensemble of 60 binary classifiers for protein localization and quantification (modified from Chong et al. 2015). Only cell images that were not filtered by the quality-control classifiers for dead cells and “ghost” objects were further classified. All cells were first classified into different cell-cycle stages using the unbudded and budded classifiers. The rest of the ensemble is organized into 20 subgroups. For some classifier groups, e.g., Cortical Patches and Cell Periphery, budded and unbudded cells were separately tested. The results from each subgroup of binary classifiers e.g., CoP, CoP-Mito, and CoP-Cyto, were consolidated with Bagging. The circles denote the percentage of 1,057,871 cells in the wild-type WT1 experiment that were assigned to each localization class, with darker green indicating a greater percentage.
Mentions: The ensLOC framework comprises several steps (Figure 1). We first segmented the cells from micrographs obtained from our wild-type screen. A total of 430 image features, including area, shape, intensity, texture, and Zernike moments (projections of image functions based on a set of orthogonal Zernike polynomials; Teague 1979) were extracted from the segmented cells. For each classifier, we used the χ2 test of independence (Liu and Setiono 1995) to select features that best discriminated the positive from the negative training instances. The filtered features were then used as input to construct the linear Support Vector Machine classifier (Platt 1998). Seventy thousand instances of cell images representative of the morphological signatures of 16 subcellular localizations were handpicked. The distribution of these training instances is shown in Figure 2 and Table 2.

Bottom Line: Changes in protein subcellular localization and abundance are central to biological regulation in eukaryotic cells.Quantitative measures of protein dynamics in vivo are therefore highly useful for elucidating specific regulatory pathways.The images depict the localization and abundance dynamics of more than 4000 proteins under two chemical treatments and in a selected mutant background.

View Article: PubMed Central - PubMed

Affiliation: The Donnelly Centre, University of Toronto, Toronto, Ontario, Canada, M5S3E1.

No MeSH data available.


Related in: MedlinePlus