Driven by the vision of high-throughput, unbiased, functional chemical screens and enabled by the functional genomics resources available for S. cerevisiae, it is now possible to screen tens of thousands of compounds for chemical-genetic interactions in just a few months’ time. To fully take advantage of this substantial increase in throughput, scalable computational tools must be developed to assist in both generating and interpreting chemical-genetic interaction profiles from these data. To this end, we developed two computational tools, BEAN-counter and CG-TARGET, the former to convert raw barcode sequencing data into chemical-genetic interaction profiles and the latter to derive actionable mode-of-action predictions from chemical-genetic interaction profiles.
BEAN-counter provides a complete toolset for processing multiplexed sequencing data from barcoded mutant pools into chemical-genetic interaction profiles. The pipeline implements several quality control and normalization steps to detect and remove technical artifacts or other systematic biases. A Python implementation of this pipeline is available at https://github.com/csbio/BEAN-counter.
CG-TARGET is a pipeline for predicting the molecular targets of compounds from chemical-genetic interaction profiles. It leverages the phenomenon that a compound’s chemical-genetic interaction profile will be similar to the genetic interaction profile of its target(s); as such, CG-TARGET uses a reference genetic interaction network to interpret the provided chemical-genetic interaction profiles. Target predictions are first made at the level of individual genes, followed by aggregation of these individual gene scores into process or pathway prediction scores in order to improve statistical confidence. An R implementation of this pipeline is available at https://github.com/csbio/CG-TARGET.
These computational toolsets have been applied to analyze more than 18,000 chemical-genetic interaction screens in S. cerevisiae. Validation experiments provide convincing evidence that CG-TARGET has the power to predict a compound’s mechanism of action at the resolution of biological processes or pathways from a single chemical-genetic interaction profile. These pipelines have enabled the functional annotation of thousands of compounds and are readily adaptable to other compound collections and mutant libraries.