Tumor Driver Identification (TDI)
Xinghua Lu, MD, PhD and Greg Cooper, MD, PhD
Tumor Driver Identification (TDI) is an algorithm that identifies genetic drivers of cancer within an individual tumor on a personalized, per-patient basis. TDI breaks current predictive barriers by implementing Bayesian causal inference methods to identify tumor drivers from genomic big data. The TDI algorithm addresses the key challenge of precision medicine in cancer — identifying which alterations are driver mutations within an individual tumor — by inferring causality from variation in genomic and transcriptional biology. Preliminary results from over 4,000 tumors show that TDI can identify known drivers as well as discover novel cancer drivers that are biologically sensible.
Phylogenetic Models For Predicting Cancer Progression
Russell Schwartz, PhD and Jian Ma, PhD
The goal of the project is to develop novel models, algorithms, and software tools to better understand the origin and evolution of tumor cells and how a patient’s tumors are likely to progress. The team brings together world-leading expertise in tumor phylogenetics, structural variant classification, and computational modeling in biology. This project leverages genome-wide sequencing of tumors in conjunction with knowledge-driven phylogenetic models and machine learning to predict what types of changes are likely to occur next. This information will contribute to personalized, precision treatment of cancer with higher likelihood of success than current approaches which do not account for the velocity of genomic changes.
Breast Cancer Recurrence Risk Predictor (PI-Predictor)
Shandong Wu, PhD
Pathology Image PI-Predictor is a computational model that combines routinely available clinical pathology test parameters (e.g. presence of hormone receptors, cellular growth markers) and radiological imaging (i.e., magnetic resonance imaging, etc.) to produce an individualized patient risk score for breast cancer recurrence. PI-Predictor uses machine learning techniques to build a computerized software model that quantitatively analyzes and integrates complimentary information from pathology tests of a biopsy sample and imaging radiomics of the entire tumor volume.
Aneurysm Prognosis Classifier (APC)
David Vorp, PhD
Aneurysm Prognosis Classifier (APC) leverages recent advances in machine learning to predict the complication risk of small abdominal aortic aneurysms (AAA). Current software tools are based solely on biomechanical properties and have not shown the ability to consistently predict aneurysm rupture. The APC algorithm will combine multiple data modalities to provide clinicians with an objective, predictive tool that can be used to guide surgical intervention decisions even before the onset of patient symptoms.
To learn more about other Alliance projects that received AWS support, click here.