Dana Farber Data Science | Cancer Research

Our Mission

We drive cancer research through innovation and collaboration in the quantitative sciences. Our faculty and staff are expert biostatisticians, computational biologists and knowledge systems engineers, unified by a commitment to improving cancer care. We strive to understand the biology of cancer, test the efficacy of treatments and extract knowledge from complex datasets. We also lead the computational groups for several collaborative cancer research centers.

Biostatistics

We are leaders in statistical methodology and clinical trial design. Our faculty provide expertise in all facets of clinical trial execution, including experimental design, logistics, and data analysis. We also analyze population and health systems data to better understand cancer risk and treatment efficacy.

Computational Biology

We build computational tools to enable basic science discoveries and improve clinical approaches in cancer and biomedical research worldwide. These include methods for processing next-generation sequencing data, statistical packages for analyzing high-throughput data, and machine learning pipelines that integrate diverse, complex datasets.

Knowledge Systems

We develop applied genomics and data sciences software to power cancer genomics research and precision cancer medicine. Our computational biologists, bioinformatics engineers, and software developers collaborate on high-impact projects in cancer genomics, immuno-oncology, clinical decision support, and data sharing.

Machine Learning

Improving diagnosis and tailoring treatments are two of the most promising ways to improve cancer outcomes. New measurement technologies provide new sources of information that can be incorporated into current best practices. We use Machine Learning tools to help investigators design new data-driven approaches to diagnosis and treatment.

Research Spotlight

Scott Carter lab looking at mathematical cancer research formulas.

ABSOLUTE

ABSOLUTE software estimates tumor purity, ploidy, and absolute copy-number inference in human cancer genomes. It has been applied to The Cancer Genome Atlas and other datasets to refine clinically interpretation of somatic cancer variants and deepen insights into tumor evolution. For example, ABSOLUTE has been used to determine whether a cancer underwent whole-genome doubling during its evolution. In addition, ABSOLUTE estimates the cancer-cell fraction of each somatic copy-number alteration and point mutation, which can be integrated to provide a rich phylogenetic description of a given cancer’s evolution.

ASK2ME

ASK2ME (All Syndromes Known 2 Man Evaluator) provides a calculator to determine cancer risk and existing management guidelines for germline mutation carriers. Using the best papers on ATM, BARD1, BRCA1, BRCA2, BRIP1, BMPR1A, CDH1, CDK4, CDKN2A, CHEK2, EPCAM, GREM1, MLH1, MSH2, MSH6, MUTYH, NBN, PALB2, PMS2, POLD1, POLE, PTEN, RAD50, RAD51C, RAD51D, SMAD4, STK11, TP53, we convert general risk estimates to patient specific ones and collate management guidelines to simplify determining a strategy.

Bioconductor

Bioconductor provides tools for the analysis and comprehension of high-throughput genomic data. Bioconductor uses the R statistical programming language and is open source/open development. The project provides cancer researchers worldwide with widespread access to a broad range of powerful statistical and graphical methods for the analysis of genomic data, including the analysis of DNA microarray, sequence, flow, SNP and other data.

Heng Li uses a whiteboard to explain concepts to his lab.

Burrows-Wheeler Aligner

The Burrows-Wheeler Aligner (BWA) is a software package for mapping DNA sequence data against a large reference genome, such as the human genome. BWA specifically targets low-divergent sequences in order to quickly analyze variation. This tool helps researchers identify genetic mutations that may be linked to cancer and other diseases. Recognized as a critical open-source tool, improvements to BWA and the launch of its next major version, BWA-MEM2, are supported by Chan-Zuckerberg Initiative Essential Open Source Software for Science funding.

Cerami lab discusses knowledge systems project for better clinical outcomes.

cBioPortal for Cancer Genomics

The cBioPortal for Cancer Genomics is an open-access, open-source resource for interactive exploration of multidimensional cancer genomics data sets. The goal of cBioPortal is to significantly lower the barriers between complex genomic data and cancer researchers by providing rapid, intuitive and high-quality access to molecular profiles and clinical attributes from large-scale cancer genomics projects, and to empower researchers to translate these rich data sets into biologic insights and clinical applications.

Cistrome

Gene regulation in humans is complex, involving thousands of proteins that regulate gene expression programs differently in hundreds of cell types. The Cistrome Data Browser is a collection of such cis-regulatory maps in human and mouse, which enables researchers to benefit from the experimental efforts of laboratories world-wide. In addition to cis-regulatory data, the Cistrome DB includes data analysis tools, such as Lisa, the Cistrome DB Toolkit and Cistrome-GO, which can help researchers find answers to questions about gene regulation, accelerating their efforts to find effective cancer treatments.

Shirley Liu explains genomic pipelines on whiteboard to student.

Immunotherapy Response Biomarkers (TIDE)

TIDE is a computational tool that evaluates tumor immune evasion as well as response and resistance to immune checkpoint blockade. Researchers derived TIDE gene expression signatures to evaluate T cell dysfunction and T exclusion in the tumor immune microenvironment. The public TIDE website demonstrates how to re-use public tumor profiles and immune-related genetic screens to prioritize immune regulators, evaluate immunotherapy response biomarkers and predict patient response to immunotherapy.

Dana-Farber clinicians discuss a patient record.

Immunotherapy Response in Melanoma

Using data from two published independent phase II melanoma clinical trials, this research project showed that Major Histocompatibility Complex (MHC) class I expression in advanced melanoma predicted resistance to anti–CTLA-4, but not anti-PD-1, treatment, which may need MHC class II to be effective. These results may explain why patients on combined therapy do better on average, with one drug overcoming the limitations of the other. The combination is also more toxic than single agents. Knowing which drug to administer to which patients could make melanoma immunotherapy less taxing without sacrificing efficacy.

Intermediate Clinical Endpoints in Prostate Cancer (ICECaP)

New systemic therapies have prolonged the lives of men with metastatic prostate cancer. The therapies may decrease mortality from prostate cancer if used when the localized disease is first diagnosed. The conduct of adjuvant prostate cancer clinical trials takes longer than a decade to reach the meaningful endpoint of overall survival. The international ICECaP Working Group is determining whether an intermediate clinical endpoint (ICE) for overall survival can be identified for future adjuvant prostate cancer clinical trials. If successful, this will expedite the design and conduct of future adjuvant therapy trials.

A slide of gene material is projected onto a screen with a silhouette figure looking at it.

Longitudinal Biomarker Screening

The incidence of hepatocellular carcinoma (HCC) in the United States continues to rise with the majority of patients diagnosed with advanced stage disease, limited treatment options and poor prognosis. However, patients with early stage disease have multiple, potentially curative treatment options making early detection an important step towards reducing the high mortality rates associated with HCC. We developed a novel biomarker screening algorithm that incorporates screening history to improve early detection for cancer.

MatchMiner

MatchMiner is an open source platform for matching patient-specific genomic profiles to precision cancer medicine clinical trials. The MatchMiner platform makes the results available in real-time to trial investigators and clinicians via a web-based platform. By rapidly analyzing a vast amount of patient genomic data, MatchMiner enables Dana-Farber clinicians to quickly identify and enroll patients in the best trials for their disease.

Parmigiani lab working at analysis on chalkboard.

Multi-Study Machine Learning

Precision medicine requires prognostic models that can perform well across different populations and clinical contexts. In machine learning, it is common for prediction models to look very promising in a training study, and disappoint in external validation. The Parmigiani lab suggests a general framework for training prediction models on multiple studies, in ways that address cross-study replicability.

Diagram showing a renewed model of breakage fusion cycles.

Discoveries in Breakage-Fusion-Bridge Cycles

A chromosome with a single broken end can undergo continuous rearrangement through a process called “breakage-fusion-bridge” cycles, first discovered in maize kernels by Barbara McClintock in the 1940s. The Zhang & Pellman labs revisited this classical phenomenon in human cells using modern techniques including live-cell imaging and whole-genome sequencing. Their analysis uncovered a cascade of new events during breakage-fusion-bridge cycles that can rapidly generate genetic variation and karyotypic complexity as seen in cancer genomes.

PROFILE

Profile is one of the nation’s most comprehensive precision cancer medicine initiatives. The project was launched as a large-scale cohort research study by scientists at Dana-Farber/Brigham and Women’s Cancer Center and Dana-Farber/Boston Children’s Cancer and Blood Disorders Center aimed at collecting as much information and genomic data as possible for every type of cancer. The long-term project goals are to determine the genetic and molecular alterations that cause cancer to develop and progress, match targeted therapies best suited for patients based on their cancer genomic profile and develop better ways to treat cancer and identify prevention strategies.

Samtools

Samtools is the reference implementation and an essential toolkit for the SAM format, which is the de facto standard to store aligned sequence data and underlies nearly all sequencing-based data analysis in cancer genomics. With over 20,000 citations, Samtools is one of the most widely used tools in bioinformatics.

Research

Our Mission

Biostatistics

Computational Biology

Knowledge Systems

Machine Learning

Research Spotlight

ABSOLUTE

ASK2ME

Bioconductor

Burrows-Wheeler Aligner

cBioPortal for Cancer Genomics

Cistrome

Immunotherapy Response Biomarkers (TIDE)

Immunotherapy Response in Melanoma

Intermediate Clinical Endpoints in Prostate Cancer (ICECaP)

Longitudinal Biomarker Screening

MatchMiner

Multi-Study Machine Learning

Discoveries in Breakage-Fusion-Bridge Cycles

PROFILE

Samtools

View Faculty

Collaborations

Get Involved

Research

Our Mission

Biostatistics

Computational Biology

Knowledge Systems

Machine Learning

Research Spotlight

Learn More About Us