Quantitative morphology of RNAi-induced phenotypes in cultured cells

National Institute On Aging

Investigators

Linked publications & trials

Abstract

We have continued to develop our high-density RNAi screening technology, and we have begun to characterize image data from small-scale pilot screens. The ability to quantify image similarity has allowed us to demonstrate that knock-down of genes known to have strong genetic or physical interactions leads to highly similar phenotypes. One of the key objectives is the ability to directly compare the phenotypes resulting from different screens. This requires the identification of critical parameters that affect reproducibility of the observed morphologies. The work is made more challenging by the sensitivity of our software, which can discern differences in morphology that are imperceptible to human observers.[unreadable] [unreadable] We have found that regulating the passage number of cells used in these experiments, optimizing growth with conditioned media, and automation of both the printing of the slides and the imaging of the experiment greatly increases reproducibility. We can now routinely generate the same results from day to day. While our software is still able to differentiate individual experiments, we are re-training our machine-based classifiers with more and more data representing a broader and broader range of the variability that we are unable to control. Once we are able to routinely and accurately classify the phenotypes in a new experiment using a classifier trained on previous data, we will begin experiments on a set of 8,000 Drosophila genes selected for their conservation through evolution.