Poster
A new benchmark dataset for phenotypic compound similarity
February 15, 2023
•
15 minute read
Phenotypic screening is a powerful technique used in drug discovery to reveal how chemical, genetic, and environmental perturbations affect cellular state. Its potential is only as strong as the tooling that enables scientists to analyze and draw conclusions.
Spring’s technologies have paved a path for multi-dimensional analysis of complex cellular phenotypes in the inflammasome pathway to empower scientists to maximize the value of high-content imaging data from a single screen.
Using this technology, our scientists have identified 7+ inflammasome inhibition targets that can be advanced in clinical programs whose pathologies are most relevant to the targets and their mechanisms of action.
Spring’s proprietary compound similarity algorithm is one of the key technologies that powered Spring’s success.
This analytical method gives scientists the ability to computationally score the phenotypic similarity between compounds. A method that can be applied to compound screening data and used as a filter to identify compounds that have true and consistent signals above experimental confounders.
We further characterized the compound similarity method by measuring the impact of inherent sources of variability such as donors, plates, and experimental runs. These measurements resolve the true biological signal of compounds over experimental noise. The characterization was performed on a bespoke library of 623 compounds, profiled in 4 batches of 8 plates each.
The accuracy of matching wells treated with the same compound provides a benchmark for the algorithm.
View our poster presented at SLAS 2023
This overall approach can be applied to other phenotypic datasets to identify similarities between compounds. Broader application of this tool can prove valuable in answering key pre-clinical questions to facilitate drug discovery efforts such as identification of MOA for novel small molecules and distinguishing novel small molecules from existing molecules in a given class.
AUTHORS
Adi Prakash, Rachel Jacobson, Francesco Rubbo, William Van Trump, Lauren Nicolaisen, Daniel Chen, Tempest Plott, Elisa Cambronero, Dat Nguyen, Christian Elabd, Ben Komalo