Stanford, CA (April 12, 2018)-- Last week, The Cancer Genome Atlas (TCGA) PanCan effort released a series of 27 papers. Published in various Cell journals, these papers mark the culmination and completion of the PanCancer Atlas Initiative (PanCan) and The Cancer Genome Atlas (TCGA) consortium. The latter represents “a comprehensive and coordinated effort to accelerate our understanding of the molecular basis of cancer through the application of genome analysis technologies, including large-scale genome sequencing.”

Faculty from Stanford’s Center for Biomedical Informatics Research (BMIR) and the Department of Biomedical Data Science (DBDS) contributed greatly to the completion of this large-scale PanCan initiative, which spanned a decade and analyzed over 11,000 tumors from 33 of the world’s most prevalent cancers; their novel methods are applied to significant effect in a number of papers, including three flagship studies published in Immunity, Cell, and Cancer Cell.

The TCGA PanCan effort brought together researchers from across the globe in a way that will continue to fuel groundbreaking research for years to come. “Participating in the PanCan project provided both opportunities and challenges,” says Dr. Andrew Gentles, Assistant Professor of Medicine (Biomedical Informatics) and, by courtesy, of Biomedical Data Science at Stanford. “It was a wonderful chance to collaborate with people across disciplines, and across continents including Tathi Malta (University of Sao Paolo) and Artem Sokolov (UCSC, now at Harvard) who drove work on deepening our understanding of the role of stemness in cancers. Of course, the logistics of coordinating these efforts are daunting. Maciej Wiznerowicz in Poland, and Ilya Shmulevich and Vesteinn Thorsson at the Institute for Systems Biology (ISB) in Seattle did a remarkable job of facilitating the immune and stemness papers. Last but not least, [PanCan] built many new scientific relationships which are proving to be the basis for new collaborations going forward.”’

The consortium’s impacts will undoubtedly be far-reaching for cancer researchers and clinical practitioners alike. “The PanCanAtlas enables for the first time the study of commonalities across cancers originating in different tissues, and can help us to move away from a paradigm based on treating cancers based on autonomy to one based on treatment based on common oncogenic processes,” explains Assistant Professor of Biomedical Informatics and Biomedical Data Science at Stanford, Dr. Olivier Gevaert.

BMIR/ DBDS Paper Highlights          

Andrew Gentles, Assistant Professor of Stanford Center for Biomedical Informatics Research and, by courtesy, of Biomedical Data Science

Aaron Newman,  Assistant Professor of Biomedical Data Science

The Immune Landscape of Cancer

For the Pan-Cancer Atlas, CIBERSORT--a computational method for studying the identities and relative amounts of distinct cell types within complex tissues (including tumors)--was used by PanCan researchers (including Dr. Gentles) to study tumor-associated immune cells. The CIBERSORT results were used directly in more than 25% of PanCan articles (7 of 27 articles), and particularly the article on the cancer immune landscape (Immunity), which explores the important roles the immune system plays in tumor growth, cancer progression, and patient outcomes, and as the target of multiple new anti-cancer therapies.

The ability of CIBERSORT to quantify immune populations was a key aspect for this Immunity paper. Some of the major findings include the discovery of “immune subtypes” of cancer that span different cancer types and represent a novel way of classifying cancers. These immune groupings were associated with the presence of specific types of mutations (variations in the tumor microenvironment) and had different survival implications for patients independent of their cancer type.

Created by Dr. Aaron Newman, Assistant Professor of Biomedical Data Science at Stanford, while he was a postdoc in Dr. Ash Alizadeh’s lab, CIBERSORT was originally published in Nature Methods in 2015. (Click here for the original press release). Since its original publication, CIBERSORT has become a widely used method for ‘in silico cytometry’, having been recently recognized as an example of “the most influential, seminal or unique reference” in its field (See Figure 1 of


Machine Learning Identifies Stemness Features Associated with Oncogenic Dedifferentiation

For this paper, molecular profiles of normal stem and progenitor like cells were used to define signatures of their properties, which could then be projected onto cancer samples. In other words, tumors could be scored for how “stem-like” they were, based on gene expression and methylation. It turned out that aspects of the immune infiltration of tumors, identified by CIBERSORT, correlated with these stemness indices, as did the expression of PD-L1, which is an important target for cancer immunotherapies.

Click here to view an animation of the stemness project, which includes an overview of TCGA.

Genomic, Pathway Network, and Immunologic Features Distinguishing Squamous Carcinomas

For this paper, Stanford’s Gevaert lab, including postdoc Dr. Kevin Brennan, a co-first author of this paper, applied their MethylMix algorithm to identify genes that are deregulated by alteration of DNA methylation, the best-known epigenetic mechanism of gene regulation, in squamous carcinomas. They identified 905 such DNA methylation-deregulated genes, including well-known cancer-related genes, such as TET1 and FANCF, which were silenced by aberrant DNA methylation specifically within the subset of SCCs occuring in the cervix and throat, that are known to be caused by human papillomavirus (HPV). This suggests that HPV may cause cancer by silencing these genes through an epigenetic mechanism.

Overall, this study used integrated analysis of various types of genome-wide data, such as genetic, epigenetic, and gene expression data, to build up a comprehensive picture of the abnormal molecular events that drive SCCs across tumor sites. This research will lead to more precise diagnosis of SCC subtypes, and highlights genes and pathways that will be investigated in future studies to identify novel drug targets.

Olivier Gevaert, Assistant Professor of Stanford Center for Biomedical Informatics Research and Biomedical Data Science

About BMIR and DBDS

The Stanford Center for Biomedical Informatics Research (BMIR) studies how acquiring and processing information can improve health and wellness, and enhance biomedical research. As a collaborative team, BMIR:

  • develops and evaluates computational methods for biomedical discovery and decision making;
  • enhances clinical care and biomedical research through information management; and
  • integrates research, training, and adoption of information technology in biomedicine.


Stanford's Department of BIomedical Data Science (DBDS) provides an intellectual home for this collaborative research, to recruit emerging talent, and to provide outstanding training to postdoctoral scholars and graduate students working in this area. As a basic science department, DBDS is devoted to the development of methods for learning from biomedical data, managing those data, and using the data to inform discovery.  Faculty create novel computational and statistical methods for acquiring, representing, storing and analyzing biological and clinical data at all scales.

Related Links

Follow the links below to read more about the PanCan TCGA consortium and the papers referenced here: