Benjamin Haibe-Kains

Picture of Dr. Benjamin Haibe-Kains

Assistant Professor

Princess Margaret Cancer Centre
Toronto Medical Discovery Tower
101 College Street, Room 11-310
Toronto, Ontario M5G 1L7

Soleil Miron, Administrative Assistant
Phone: +1 (416) 581-7628
Email Dr. Benjamin Haibe-Kains

At A Glance:

  • Focus of the lab includes Computational Biology, Bioinformatics, Machine learning, Cancer Genomics and Pharmacogenomics
  • Strong emphasis on translational research, with close collaboration with clinicians to develop relevant biomarkers and novel therapeutic strategies
  • Large network of collaborators in multiple cancer types
  • Integrative analysis of multi -omics data in different model systems (cell lines, xenografts, primary tumors)
  • Algorithmic and software development at the core of the lab research
  • We make our research fully reproducible!


Short Bio:

Benjamin Haibe-Kains is Scientist at the Princess Margaret Cancer Centre (PM), University Health Network, and Assistant Professor in the Medical Biophysics and Computer Science departments of the University of Toronto. Dr. Haibe-Kains earned his PhD in Bioinformatics at the Université Libre de Bruxelles (Belgium), for which he was awarded a Solvay Award (Belgium). Supported by a Fulbright Award, Dr. Haibe-Kains did his postdoctoral fellowship at the Dana-farber Cancer Institute and Harvard School of Public Health (USA). Dr. Haibe-Kains started his own laboratory at the Institut de Recherches Cliniques de Montréal (Canada) and moved to PM in November 2013. Dr. Haibe-Kains’ research focuses on the integration of high-throughput data from various sources to simultaneously analyze multiple facets of carcinogenesis. Dr. Haibe-Kains and his team are analyzing high-throughput (pharmaco)genomic datasets to develop new prognostic and predictive models and to discover new therapeutic regimens in order to significantly improve disease management. Dr. Haibe-Kains' main scientific contributions include several prognostic gene signatures in breast cancer, subtype classification models for ovarian and breast cancers, as well as genomic predictors of drug response in cancer cell lines.


Major Contributions:

Predictors of drug response based on pharmacogenomic data. My laboratory is focusing on developing robust genomic predictors of drug response from pharmacogenomic data. The Cancer Genome Project (CGP) and the Cancer Cell Line Encycolpedia (CCLE) studies recently published drug sensitivity data on large panel of genomically-characterized cancer cell lines with the aim of unraveling new associations between genomic features of these cell lines and their response to drugs. We used CGP data to train genomic predictors of response to 15 drugs screened in both studies, and showed that half of these models could not be validated on CCLE (J Am Med Inform Assoc 2013). We then investigated the consistency between these two large pharmacogenomic studies as potential cause of failure for most of our predictors and discovered that, although gene expression data were highly concordant, the drugs sensitivity data were highly inconsistent across studies (Nature 2013; Cancer Res 2014). We are now developing new analysis pipelines to increase robustness of drug phenotypic measurements to build more robust genomic predictors of drug response.

Large­-scale causal gene regulatory networks. With the recent interest in network medicine I decided to investigate the use of gene interactions extracted from the biomedical literature and structured biological databases (referred to as “priors”) to better infer gene-gene interaction networks from gene expression data. I implemented predictionet, a network inference approach integrating genomic data and priors, and developed a new validation framework based on high-throughput perturbation experiments (Genomics 2014). Collaborating with Entagen, I have implemented a web application ( that enable users to easily infer predictive gene interaction networks by combining interactions extracted from publications, pathway databases and gene expression data (Nucleic Acids Res 2012).

Gene expression-based classification models to robustly identify cancer molecular subtypes. Molecular subtyping of cancers is key in the path of personalized medicine. However the initial cancer subtyping models (I) were not robust when applied to independent datasets; and (II) were based on the expression of large numbers of genes, which complicated their translation into clinical assays. These shortcomings prompted me to develop a simple, yet robust, three-gene classification model for breast cancer molecular subtyping. I showed that this model was more robust than published models while yielding similar prognostic value than more complex gene signatures (J Natl Cancer Inst 2012); I am now working with Caprion Inc. to translate this model into a clinical assay. I have applied a similar approach in ovarian cancer and identified a new angiogenic subtype with significant prognostic value (PLoS ONE 2012). I have developed a novel trans-species classifier of meduloblastoma molecular subtypes (Genomics 2015).

Prognostic biological processes in breast and ovarian cancer subtypes. As numerous microarray datasets are publicly available, I quickly recognize the need for a bioinformatics framework that enables joint analysis of gene expression data generated from different platforms and laboratories. I designed a meta-analysis framework to robustly quantify activity of hallmarks in breast cancer (estrogen and her2 signaling pathways, proliferation, angiogenesis, apoptosis, immune response, and tumor invasion), and assessed their prognostic value in each molecular subtype separately. This framework opened the way for many other studies investigating subtype-specific biomedical questions (Clin Cancer Res 2008).  Two of the signatures we have discovered have been patented. I have recently extended my meta-analysis framework to ovarian cancer (PLoS Comput Biol 2013).


Selected References:

Link to Pubmed Publications
Link to Google Scholar

  • Inconsistency in large pharmacogenomic studies. Haibe-Kains B, El-Hachem N, Birkbak NJ, Jin AC, Beck AH, Aerts HJ, Quackenbush J. Nature. 2013 Nov 27. doi: 10.1038/nature12831. PMID: 24284626

  • mRMRe: an R package for parallelized mRMR ensemble feature selection. De Jay N, Papillon-Cavanagh S, Olsen C, El-Hachem N, Bontempi G, Haibe-Kains B. Bioinformatics. 2013 Jul 3 PMID: 23825369

  • Comparison and validation of genomic predictors for anticancer drug sensitivity. Papillon-Cavanagh S, De Jay N, Hachem N, Olsen C, Bontempi G, Aerts HJ, Quackenbush J, Haibe-Kains B. J Am Med Inform Assoc. 2013 Jan 26. PMID: 23355484

  • Significance Analysis of Prognostic Signatures. Beck AH, Knoblauch NW, Hefti MM, Kaplan J, Schnitt SJ, Culhane AC, Schroeder MS, Risch T, Quackenbush J, Haibe-Kains B. PLoS Comput Biol. 2013 Jan;9(1):e1002875. PMID: 23365551

  • A three-gene model to robustly identify breast cancer molecular subtypes. Haibe-Kains B, Desmedt C, Loi S, Culhane AC, Bontempi G, Quackenbush J, Sotiriou C. J Natl Cancer Inst. 2012 Feb 22;104(4):311-25. PMID: 22262870

  • Predictive networks: a flexible, open source, web application for integration and analysis of human gene networks. Haibe-Kains B, Olsen C, Djebbari A, Bontempi G, Correll M, Bouton C, Quackenbush J. Nucleic Acids Res. 2011 Nov 16. PMID: 22096235

  • PIK3CA mutations associated with gene signature of low mTORC1 signaling and better outcomes in estrogen receptor-positive breast cancer. Loi S, Haibe-Kains B, Majjaj S, Lallemand F, Durbecq V, Larsimont D, Gonzalez-Angulo AM, Pusztai L, Symmans WF, Bardelli A, Ellis P, Tutt AN, Gillett CE, Hennessy BT, Mills GB, Phillips WA, Piccart MJ, Speed TP, McArthur GA, Sotiriou C. Proc Natl Acad Sci U S A. 2010 Jun 1;107(22):10208-13. doi: 10.1073/pnas.0907011107. Epub 2010 May 17. PMID: 20479250

  • A fuzzy gene expression-based computational approach improves breast cancer prognostication. Haibe-Kains B, Desmedt C, Rothé F, Piccart M, Sotiriou C, Bontempi G. Genome Biol. 2010;11(2):R18. Epub 2010 Feb 15. PMID: 20156340