NCI U01 CA-202144: Multi-Tensor Decompositions for Personalized Cancer Diagnostics and Prognostics

In the Genomic Signal Processing Lab, we invented the "eigengene,"1,2,3  and pioneered the matrix4,5,6  and tensor7,8,9  modeling of molecular biological data, which, as we demonstrated, can (i) compare and integrate different, e.g., multi-omic, data types; (ii) scale with data sizes; (iii) be interpreted in terms of the known biology and batch effects underlying the data; and (iv) correctly predict previously unknown and experimentally verifiable physical,10,11,12  cellular,13,14,15,16  and evolutionary17,18  mechanisms.19,20,21  Supported by a five-year, three and a half million-dollar National Cancer Institute (NCI) Physical Sciences in Oncology project grant,22,23  our successful retrospective clinical trial24,25  provided a proof of principle that our recently formulated and developed physics-inspired multi-tensor26,27,28,29  generalizations of the singular value decomposition additionally can (v) discover accurate, precise, and actionable genotype-phenotype relationships; (vi) prove relevant to populations based upon whole genomes of small cohorts; and (vii) be validated in clinical trials.

Our trial experimentally validated a genome-wide pattern of DNA copy-number alterations found in glioblastoma brain30,31,32,33,34,35  tumors as a predictor of life expectancy statistically more accurate than and independent of age, which has been the best indicator for 70 years. We discovered this pattern, and patterns similarly predicting survival and response to treatment of patients with neuroblastoma nerve cancer and lung,36,37  ovarian,38,39,40,41,42,43,44  and uterine adenocarcinoma cancers, by modeling data from NCI databases. Researchers have recognized copy-number alterations as a hallmark of cancer for more than a century and have observed them in these tumors for decades. However, previous attempts to associate the tumors' copy numbers with the patients' outcome failed, including studies of the same data that used other artificial intelligence and machine learning methods. This demonstrates that our multi-tensor decompositions are uniquely suited for personalized medicine.

    Recent Research in the News

  1. Press Release: J. Kiefer, "Genome-Wide Pattern Found in Tumors from Brain Cancer Patients Predicts Life Expectancy," American Association for the Advancement of Science (AAAS) EurekAlert! (May 15, 2020).
  2. Feature: A. J. Engler and D. E. Discher, "Rationally Engineered Advances in Cancer Research," Applied Physics Letters (APL) Bioengineering 2 (3), Special Topic: Bioengineering of Cancer preface 031601 (September 2018).
  3. Abstract: O. Alter, "Multi-Tensor Decompositions for Personalized Cancer Diagnostics and Prognostics," Physical Sciences in Oncology Network (PS-ON) of the Cancer Research Institute (NCI) (September 2015).1
  4. Feature: F. Pavlou, "Big Data, Hidden Knowledge," The Pathologist (June 15, 2015).2
  5. Feature: R. Atkins, "Calculating Cancer Cures," National Academy of Engineering (NAE) Innovation Podcast and Radio Series (April 19, 2015).3
  6. Press Release: J. Kiefer, "New Method Increases Accuracy of Ovarian Cancer Prognosis and Diagnosis," American Association for the Advancement of Science (AAAS) EurekAlert! (April 15, 2015).
  7. Review: M. Méchali, Faculty of 1000 evaluation 1728974 (February 2010).
  8. Feature: S. N. Dwivedi, "Rao Conference at the Interface between Statistics and the Sciences (Hyderabad, India, December 30, 2009 – January 2, 2010), Rao Best Poster Prize," International Biometric Society (IBS) Bulletin 27 (1), pp. 6–7 (January–March 2010).
  9. Synopsis: O. Alter, "2005 Linear Algebra and its Applications (LAA) Lecture," IMAGE: International Linear Algebra Society (ILAS) Bulletin 35, pp. 2–15 (December 2005).4
  10. Feature: M. E. Kilmer and C. D. Moravitz Martin, "Decomposing a Tensor," Society for Industrial and Applied Mathematics (SIAM) News 37 (9), (November 2004).
  11. Feature: J. Wixon and J. Ashurst, "Genome Informatics," Computational Functional Genomics 4 (5), pp. 509–514 (October 2003).
  12. Invited Commentary: L. Y. Dirix and A. T. van Oosterom, "Gene-Expression Profiling to Classify Soft-Tissue Sarcomas," Lancet 359 (9314), pp. 1263–1264 (April 2002).
  13. Feature: B. H. Ripin, "1998 Outstanding Doctoral Thesis Research in Atomic, Molecular, or Optical Physics (DAMOP) Award Finalists," American Physical Society (APS) News 7 (8), p. 5 (August–September 1998).