Supervised ML - SAMP-Score

Machine learning classifier for identifying novel pro-senescence compounds in p16-positive cancers

By Ryan J Wallis, PhD

I developed a machine learning model for the classification of senescent cells based on their cell morphology - SAMP-Score.

SAMP-Score was built and validated through analysis of high-throughput screening data and multiparameter phenotypic analysis of the following large scale data sets:

  • Genome-wide siRNA (x63 384-well plates – 24,122 treatments).
  • Compound diversity library (x61 384-well plates – 20,000 treatments).
  • Hit validation dose response (x30 384-well plates – 8,940 treatments).

SAMP-Score has been used to identify novel compounds that initiate a senescence response in cancer cells, demonstrating its utility as a drug discovery tool.

ML Techniques: Regression (Logistic), regularisation (Lasso, Elastic Net), support vector machines (SVM), decision trees (Random Forrest), dimensionality reduction (MDA), neural networks, ensemble model stacking and hierarchical clustering

Go to the Pre-print

Github Link Embargoed Until Publication

Share: Twitter LinkedIn