Learn more →
Back to Expert Scholars
research / researchpharmacogenomics, ORCESTRA, reproducibility in cancer AI, transcriptional signatures, open scienceBispecific T-cell Engager Pioneer

Benjamin Haibe-Kains

本杰明·海贝-凯恩斯

PhD

🏢Princess Margaret Cancer Centre, University Health Network; University of Toronto(加拿大玛格丽特公主癌症中心,大学健康网络;多伦多大学)🌐Canada

Senior Scientist, Princess Margaret Cancer Centre; Professor of Medical Biophysics, University of Toronto; Head, Bioinformatics and Computational Genomics Laboratory玛格丽特公主癌症中心高级科学家;多伦多大学医学生物物理学教授;生物信息学与计算基因组学实验室负责人

55
h-index
4
Key Papers
5
Awards
4
Key Contributions

👥Biography 个人简介

Benjamin Haibe-Kains, PhD is Senior Scientist at Princess Margaret Cancer Centre (University Health Network) and Professor of Medical Biophysics at the University of Toronto, where he heads the Bioinformatics and Computational Genomics Laboratory. He is internationally recognized for his work in pharmacogenomics, cancer biomarker validation, and scientific reproducibility in computational oncology. Dr. Haibe-Kains pioneered large-scale computational studies of cancer cell line pharmacogenomics, analyzing datasets from the Cancer Cell Line Encyclopedia (CCLE) and Genomics of Drug Sensitivity in Cancer (GDSC) to develop machine learning models predicting drug sensitivity from genomic features. He co-authored landmark papers in Nature demonstrating profound inconsistencies in published pharmacogenomic studies, exposing data processing disparities, batch effects, and analytical choices that led to contradictory conclusions — catalyzing a major debate about reproducibility in computational cancer research. In response, he developed ORCESTRA (Orchestrating and Sharing Translational Research with Automated Pipelines), a computational platform for creating, sharing, and reproducing cancer pharmacogenomic data analyses with full transparency and version control. Dr. Haibe-Kains is also recognized for developing genefu, an R package for computing and validating breast cancer molecular subtype signatures (PAM50, Oncotype DX, MammaPrint), which has become a standard tool in breast cancer bioinformatics. His laboratory has consistently championed open-source software, open data, and reproducible analytical standards in computational oncology.

Share:

🧪Research Fields 研究领域

Pharmacogenomics — AI Models Predicting Drug Sensitivity from Cancer Cell Line and Patient Genomics药物基因组学——从癌症细胞系和患者基因组预测药物敏感性的AI模型
ORCESTRA — Reproducible and Transparent Cancer Biomarker Research PlatformORCESTRA——可重现和透明的癌症生物标志物研究平台
Transcriptional Signature Validation — Reproducibility Assessment of Genomic Prognostic Signatures转录特征验证——基因组预后特征的可重现性评估
Open Science in Cancer AI — Code Sharing, Data Standards, and Reproducible Research癌症AI的开放科学——代码共享、数据标准和可重现研究
Multi-Omics Drug Response Models — Integrating Genomics, Transcriptomics, and Epigenomics for Precision Oncology多组学药物反应模型——整合基因组学、转录组学和表观基因组学用于精准肿瘤学

🎓Key Contributions 主要贡献

Pharmacogenomics Reproducibility — Exposing Inconsistencies in Cell Line Drug Sensitivity Studies

Published landmark papers in Nature (2013) and Nature Medicine identifying profound discordance between major pharmacogenomic datasets (CCLE and GDSC) in drug sensitivity measurements and genomic predictors, demonstrating that differences in data processing, cell line nomenclature, and normalization methods led to contradictory drug sensitivity predictions and biomarker associations. This work triggered a major scientific debate about reproducibility standards in computational cancer pharmacology and led to coordinated efforts to standardize cell line genomics and drug sensitivity data across consortia.

ORCESTRA — Reproducible Pharmacogenomics and Cancer Biomarker Analysis Platform

Developed ORCESTRA (Orchestrating and Sharing Translational Research with Automated Pipelines), a web-based computational platform that provides reproducible, version-controlled, and transparently documented cancer pharmacogenomics data analysis pipelines. ORCESTRA enables researchers to create, share, and reproduce pharmacogenomic data objects (PharmacoSets) that contain harmonized drug sensitivity data, genomic profiles, and metadata with full analytical provenance, addressing the reproducibility crisis in computational pharmacogenomics and establishing a model for transparent cancer bioinformatics.

Breast Cancer Molecular Signature Computation — genefu R Package

Developed and maintains genefu, a widely-used R/Bioconductor package that implements computational algorithms for breast cancer molecular subtyping (PAM50, claudin-low, AIMS), prognostic signature scoring (Oncotype DX simulation, MammaPrint simulation, Genomic Grade Index), and survival analysis from gene expression data. With tens of thousands of downloads, genefu has become a standard tool in breast cancer bioinformatics used in hundreds of published studies for standardized molecular subtype assignment and signature validation.

Machine Learning for Drug Sensitivity Prediction from Multi-Omics

Developed and benchmarked machine learning algorithms for predicting cancer drug sensitivity from multi-omics features (gene expression, copy number, mutation, methylation) using large-scale pharmacogenomic training datasets. Contributed methodological advances in feature selection, cross-validation design, and independent validation frameworks for pharmacogenomics ML models, and published systematic comparisons demonstrating that transcriptomic features outperform genomic mutation profiles for predicting sensitivity to most drugs across cancer cell line pharmacogenomics datasets.

Representative Works 代表性著作

[1]

Inconsistency in Large Pharmacogenomic Studies

Nature (2013)

Landmark analysis revealing profound inconsistencies between CCLE and GDSC pharmacogenomics datasets, exposing data processing and methodological issues that undermined reproducibility of drug sensitivity biomarker studies.

[2]

Towards Reproducibility in Big Cancer Data

Nature Cancer (2020)

Framework for reproducible cancer data science covering data sharing standards, analytical provenance, software packaging, and code review practices for computational oncology studies.

[3]

genefu: An R/Bioconductor Package for Computation of Gene Expression-Based Signatures in Breast Cancer

Bioinformatics (2016)

Description and validation of genefu R package for breast cancer molecular subtype computation and prognostic signature scoring from gene expression data.

[4]

ORCESTRA: Enabling Reproducible Biomarker Research in Pharmacogenomics

npj Precision Oncology (2021)

Description of the ORCESTRA platform for reproducible pharmacogenomics data analysis with automated pipelines and versioned data objects for transparent cancer biomarker research.

🏆Awards & Recognition 奖项与荣誉

🏆Princess Margaret Cancer Research Award for Computational Biology
🏆University of Toronto Excellence in Research Award
🏆Canadian Institutes of Health Research (CIHR) New Investigator Award
🏆Bioinformatics Open Source Award — Outstanding Contribution
🏆AACR Scholar-in-Training Award for Computational Oncology

📄Data Sources 数据来源

Last updated: 2026-04-06 | All information from publicly available academic sources

关注 本杰明·海贝-凯恩斯 的研究动态

Follow Benjamin Haibe-Kains's research updates

留下邮箱,当我们发布与 Benjamin Haibe-Kains(Princess Margaret Cancer Centre, University Health Network; University of Toronto)相关的新研究或访谈时,我们会通知你。

我们不会泄露你的信息,也不会发送无关内容。随时可以退订。

Explore More Experts

Discover the researchers shaping the future of cancer treatment