Category: Data Analysis and Informatics

1308-D - Facilitating the Analysis of High-Dimensional Quantitative High-Throughput Screening Data: Profiling cancer cell lines for drug repurposing opportunities.

Tuesday, February 6, 2018
5:00 PM - 6:00 PM

Quantitative high throughput screening (qHTS) allows the investigation of a wide range of compound potencies while the dose response curves provide the metrics that help validate the results. One of the challenges of interrogating qHTS dose response data, beyond the sheer volume of data, is that hit selection requires the simultaneous assessment of multiple factors, including comparing potency, efficacy, and quality of curve fits across each of the various readouts and counter screens for a screening campaign including cytotoxicity and assay artifacts. The goal is to quickly make compound selections from complex data sets, prioritizing the best hits for further study. We have met these challenges by the design of specialized analysis workflows or templates in Spotfire that work with our qHTS database systems to address a variety of analysis scenarios. In parallel, we have instituted an internal training program to enable scientists to effectively analyze their data sets and distribute the analysis effort across the team to make critical research decisions. Recent results on the discovery of drug repurposing opportunities for pediatric cancer provides an illustration of how the analysis is performed. We ran qHTS on a collection of 20 cell lines, one non-cancer cell line and 19 cell lines derived from pediatric solid tumors. These cell lines were screened against an 11-point titration of a collection of 4,728 substances (3,886 distinct compounds) comprised of approved drugs and investigational compounds. The ability to cluster compound response profiles across this panel of cell lines coupled with interactive visualization of dose-response curves streamlined our compound selection process. The ability to overlay compound annotations and cluster cell lines by response profile assisted with finding high-quality compounds effective against all tested cancer cell lines as well as compounds effective against specific cancer types while having minimal effect on our non-cancer cell line.

John C. Braisted

Senior Software Engineer
NIH/NCATS
Rockville, MD

I have 15 years experience working in physiology and biochemistry labs as a research assistant. 15 years experience working in informatics. Degrees in Zoology and Computer Science.