Skip to main content

Table 1 Average classification accuracies using TAS statistics on Endogenous and Transfected data test sets and the subcellular markers used

From: Fast automated cell phenotype image classification

Organelle

Endogenous Accuracy

Transfected Accuracy

Endogenous Marker

Transfected Marker

Nucleus

97.4%

92.6%

DAPI

myc-nbp-45 (txreg I920050F21)

Cytoplasm

-

95.5%

-

GFP-V5

Endoplasmic Reticulum

90.8%

82.5%

anti-PDI

ICAT-GFP

Golgi

98.2%

87.7%

anti-Beta-COP

GCC-GFP

Plasma Membrane

93.3%

82.4%

anti-EGFR

myc-Lysophosphatidic acid receptor

Endosome

93.9%

88.1%

anti-SNX1

rab5a-GFP

Lysosome

96.5%

76.5%

Lysotracker

myc-chloride channel 7 (5330412O18)

Peroxisome

95.9%

95.7%

anti-catalase

ALD

Mitochondria

92.6%

76.2%

Mitotracker

myc-carntine/acylcarnitine translocase

Actin Cytoskeleton

97.8%

92.3%

Phalloidin

YFP-actin

Microtubules

92.8%

92.1%

anti alpha-tubulin

YFP-tubulin

  1. The accuracies (percentage of true positive predictions) are averaged over 1000 random division of each data set into 4/5ths for training and 1/5th for testing. The overall average classification accuracies on the test sets were 94.4% and 86.3% on the endogenous and transfected sets, respectively.