RESEARCH

The long-term objective of the Ideker Lab is to create artificially intelligent, mechanistic models of cancer and neurodegenerative diseases for translation of patient data to precision diagnosis and treatment. We seek to advance this goal by addressing fundamental questions in the field: What are the genetic and molecular networks that promote disease, and how do we best chart these? How do we use knowledge of these networks in intelligent systems for predicting the effects of genotype on phenotype? In what follows, we will discuss our five most significant areas of research with attention to recent advancements.

Mapping of subcellular networks and structures
Visible machine learning for predictive oncology
Network biomarkers for cancer and neuropsychiatric disease
Epigenetic Aging
Bioinformatics Software

Mapping of genetic and molecular networks across species, diseases and cell states

Our lab has long promoted strategies to experimentally map and analyze the molecular networks that encode biological function. This interest began during my PhD work, where Leroy Hood and our team outlined the “Systems Biology” approach to studying biological systems by “perturbing them systematically; monitoring the global response at multiple levels (gene, protein, metabolite); and formulating network models that describe the structure of the system and its response to perturbation” [a]. Network mapping has continually driven the research agenda of our laboratory through our interest in complementary approaches, particularly in mapping protein interaction networks with affinity purification mass spectrometry or yeast-two-hybrid [b], as well as mapping epistatic genetic interactions by combinatorial gene knockout [c,d]. Recently, we have contributed to large team efforts to map the physical interaction landscapes of multiple cancer types and SARS-CoV-2. We are also making significant progress in integrating protein interaction networks with protein immunofluorescent images to reconstruct most human cell components and chart new ones.

  1. Ideker T, et al. Integrated genomic and proteomic analyses of a systematically perturbed metabolic network. Science. 2001 May 4;292(5518):929-34. PMID: 11340206.
  2. Suthram S, et al. The Plasmodium protein network diverges from those of other eukaryotes. Nature. 2005 Nov 3;438(7064):108-12. PMCID: PMC2830740.
  3. Bandyopadhyay S, et al. Rewiring of genetic networks in response to DNA damage. Science. 2010 Dec 3;330(6009):1385-9. PMCID: PMC3006187.
  4. Shen JP, et al. Combinatorial CRISPR-Cas9 screens for de novo mapping of genetic interactions. Nature Methods. 2017 Mar 20. PMID: 28319113. PMCID PMC5449203.

Cover Article: Translation of Genotype to Phenotype by a Hierarchy of Cell Subsystems

Significant Publication: Multi-Scale Integrated Cell (MuSIC) Maps chart eukaryotic cell architecture by systematically integrating proteomic data modalities across scales.

Qin Y, …, Lundberg E*, Ideker T*. Mapping cell structure across scales by fusing protein images and interactions. Nature. 2021 Nov 24. PMID: 34819669 *Co-corresponding

Visible machine learning for predictive oncology and other genotype-phenotype challenges

We are developing “visible” machine learning approaches to model the flow of genetic information, in which predictive models learn not only to translate genotype to phenotype but to also identify the molecular functions and mechanisms by which these predictions are made. The central concept is to couple the structure of a machine learning model to the structure and function of a target biological system, creating what we have called visible neural networks (VNNs). Whereas a typical predictive model is not readily interpretable due to many hidden variables and states (a “black box”), VNNs can be more directly inspected to reveal the molecular and cellular events responsible for each prediction. They provide a framework for interpretable deep learning, combining the generality, scale and power of neural networks with a biomechanistic understanding. These concepts led to DCell, a deep neural network modeling approximately 3500 subsystems in a budding yeast cell and which is able to accurately translate genotypes to growth phenotypes. Using DCell as a foundation, we created deep neural networks of cancer which predict response to therapy based on the tumor genotype (i.e. its profile of genetic markers) and the drug formula. Visible deep learning models open up the possibility to address a host of biomedical questions which have been recalcitrant to machine learning thus far, with significant applications in cancer genomics. 

  1. Ma J, Yu M, Fong S, et al. Using deep learning to model the hierarchical structure and function of a cell. Nature Methods. 2018 Mar 5. doi: 10.1038/nmeth.4627
  2. Michael KY, et al. Visible machine learning for biomedicine. Cell. 2018 Jun 14;173(7):1562-5. PMCID: PMC6483071
  3. Ma J, et al. Few-shot learning creates predictive models of drug response that translate from high- throughput screens to individual patients. Nature Cancer. 2021 Feb;2(2):233-44. DOI: 10.1038/s43018-020-00169-2.
  4. Zhao X, Singhal A, Park S, Kong J, Bachelder R, Ideker T. Cancer mutations converge on a collection of protein assemblies to predict resistance to replication stress. Cancer Discovery. 2024 Mar 1;14(3):508-23.

Cover Article: Using deep learning to model the hierarchical structure and function of a cell

Network biomarkers for cancer and neuropsychiatric disease

Complex genetic diseases like cancer often involve multiple subtypes with distinct causes and clinical outcomes. Genome sequences offer a rich source of data for recognizing and classifying these subtypes in patient populations. However, comparing genomes has proven challenging since two patients rarely share the same mutations. To address this challenge, our laboratory developed the concept of network biomarkers in 2007 [1a below], which integrates heterogeneous coding and non-coding alterations through common signaling and transcriptional networks. The key idea behind this approach is that complex diseases are difficult to analyze because they can invoke different genetic causes in different patients, but these causes often converge at levels of organization higher than individual genes, as captured by molecular networks.

We initially applied the network biomarkers approach to classify patient subtypes in chronic lymphocytic leukemia, successfully translating it to the clinic in 2012. This success motivated our continued work in advancing tools for identifying biomarkers as networks rather than individual genes or proteins. Notable subsequent work includes the development of network-based stratification (NBS) [1b], a method to stratify tumor populations into informative subtypes by clustering patients with mutations in common network regions. Collectively, these efforts have led to further studies by us [1c,d] and many other research groups, who have refined the methods or networks underlying network-based biomarkers across a variety of diseases.

  1. Chuang HY, et al. Network-based classification of breast cancer metastasis. Mol Syst Biol. 2007 Nov;10(11):1108-15. PMCID: PMC2063581.
  2. Hofree M, et al. Network-based stratification of tumor mutations. Nature Methods 2013 10(11):1108-15. PMCID: PMC3866081.
  3. Gross AM, et al. Multi-tiered genomic analysis of head and neck cancer ties TP53 mutation to 3p loss. Nature Genetics. 2014 Sep;46(9):939-43. PMCID: PMC4146706.
  4. van de Haar J, et al. Identifying epistasis in cancer genomes: a delicate affair. Cell. 2019 May 30;177(6):1375-83. PMCID: PMC6816465.

Cover Article: A gene ontology inferred from molecular networks

Significant Publications: Interpretation of cancer mutations using a multiscale map of protein systems

Zheng F†, Kelly M†, …, Fraley S, Gutkind S, Krogan N*, Ideker T*. Interpretation of cancer mutations using a multiscale map of protein systems. Science. 2021 374 (6563). PMID: 34591613. †Co-first, *Co-corresponding

Significant Publications: Predicting drug response and synergy using a deep learning model of human cancer cells

Kuenzi B†, Park J†, Fong S, …., Ideker T. Predicting drug response and synergy using a deep learning model of human cancer cells. Cancer Cell. 2020 Nov 9;38(5):672-684.e6. PMID:33096023. †Co-first

Epigenetic Aging 

In 2012 we showed that large parts of the methylome are remodeled with age, a process that is accelerated by disease and slowed in certain genotypes and in women versus men. These findings led to the “epigenetic clock” model for predicting rate of biological aging [a]. We have since reported that these changes are accelerated by viral infection [b] and slowed by anti-aging treatments such as caloric restriction and rapamycin [c]. Most recently, we used epigenetic profiles to translate age between humans and dogs [d]. Comparison of Labrador retriever and human methylomes revealed a nonlinear relationship between dog and human aging which did not follow the conventional wisdom that 1 dog year = 7 human years, leading to a story that was popularized by many news outlets.

    1. Hannum G, et al. Genome-wide methylation profiles reveal quantitative views of human aging rates. Molecular Cell. 2013 Jan 24;49(2):359-67. PMCID: PMC3780611.
    2. Gross AM, et al. Methylome-wide analysis of chronic HIV infection reveals five-year increase in biological age and epigenetic targeting of HLA. Molecular Cell. 2016 Apr 21;62(2):157-68. PMCID: PMC4995115.
    3. Wang T, et al. Epigenetic aging signatures in mice livers are slowed by dwarfism, calorie restriction and rapamycin treatment. Genome Biology. 2017 Dec;18(1):1-1. PMCID: PMC5371228.
    4. Wang T, et al. Quantitative translation of dog-to-human aging by conserved remodeling of the DNA methylome. Cell Systems. 2020 Aug 26;11(2):176-85. PMCID: PMC7484147.

      Cover Article: Quantitative Translation of Dog-to-Human Aging by Conserved Remodeling of the DNA Methylome

      Bioinformatics Software 

      Our lab is involved in development of bioinformatic resources that are widely used in biomedical research. The most visible of these is the network analysis platform Cytoscape (www.cytoscape.org). It is a principal tool used by researchers to create and visualize models of molecular interaction networks, with approximately 20,000 downloads per month and >40,000 citations to the original Cytoscape marker paper. More than half of these citations have been added in the past five years, underscoring the continued relevance of the software and propelling the marker paper to the status of most highly cited work in the journal Genome Research. The platform includes an appstore for third-party Cytoscape analysis tools, with >350 such plugins or “Apps” and a cloud-based storage system for networks. Finally, we have been working on significant new Cytoscape functionality for detecting communities of proteins (in protein-protein interaction networks) or cells (in single-cell RNA sequencing data).

      1. Shannon P, et al. Cytoscape: A software environment for integrated models of biomolecular interaction networks. Genome Research. 2003 Nov;13(11):2498-504. PMCID: PMC403769..
      2. Saito R, et al. A travel guide to Cytoscape plugins. Nature Methods. 2012 Nov;9(11):1069-76. PMCID: PMC3649846.
      3. Pratt D, et al. NDEx: The Network Data Exchange. Cell Systems. 2015 Oct 28;1(4):302-305. PMCID: PMC4649937
      4. Zheng F, et al. HiDeF: identifying persistent structures in multiscale ‘omics data. Genome Biology. 2021 Dec;22(1):1-5. PMCID: PMC7789082.

      Research Focus Keywords: Systems Biology, Computational Biology & Data Science, Genomics, Machine Learning, Integrative Biology, Cancer, Neuropsychiatric disease, Cell Mapping