Case Study: Rosetta Inpharmatics achieves accurate breast cancer outcome prediction with MathWorks MATLAB

Rosetta Predicts the Clinical Outcome of Breast Cancer Patients

Rosetta Inpharmatics, a Merck & Company subsidiary, partnered with the Netherlands Cancer Institute to tackle a difficult clinical problem: predicting which breast cancer patients will develop distant metastases so clinicians can avoid unnecessary chemotherapy. The team needed software that combined powerful statistics, large-scale data handling, and image processing for TIFF microarray slides, and they chose MathWorks tools—specifically MATLAB and the Image Processing Toolbox—to accelerate custom analysis and prototyping.

Using MathWorks’ MATLAB platform and toolboxes, Rosetta Inpharmatics extracted signal intensities from microarray images, performed DNA microarray analysis, developed hierarchical clustering, and built a classifier that identified a 70‑gene signature predictive of outcome from a 117‑patient study (followed >5 years). The MathWorks‑based solution produced accurate prognoses, reduced estimated unnecessary toxic chemotherapy from about 90% to nearly 40%, sped scientist feedback and prototyping (including for the commercial Rosetta Resolver product), and offered flexibility for rapid algorithm adjustments.

Rosetta Inpharmatics

Hongyue Dai

Director of Custom Analysis