Researchers at the GW Cancer Center are combining artificial intelligence, genomics, and large-scale biomedical data to uncover how cancer begins. And how treatments can be better tailored to each patient.
Understanding why cells change — and why some become cancerous — is one of the most complex questions in modern medicine. At the George Washington University Cancer Center, researchers Anelia Horvath, PhD, MSc, and Raja Mazumder, PhD, MS, MSc, are using artificial intelligence to analyze genetic data at an unprecedented scale, revealing molecular patterns that could help doctors diagnose disease earlier and select more effective treatments.
Their work sits at the intersection of genomics, computational biology, and precision medicine, fields that are rapidly transforming how scientists study cancer.
Seeing What Traditional Research Can Miss
Cells regulate thousands of genes simultaneously. Even subtle shifts in how these genes are activated can influence whether a cell remains healthy or begins to behave abnormally.
Horvath studies these changes using advanced computational tools that analyze DNA and RNA data from individual cells. RNA, the molecule that helps carry out genetic instructions, plays a key role in determining how cells function. When RNA activity changes, it can alter how genes behave — and potentially how diseases develop.
Artificial intelligence allows researchers to examine these signals across enormous genomic datasets to mine the literature, and identify patterns that would be nearly impossible to detect through traditional analysis.
Connecting Data Through the FEAST Project
Horvath and Mazumder collaborate with researchers worldwide through an international initiative known as Federated Ecosystems for Analytics and Standardized Technologies (FEAST). The project is developing a novel federated architecture for analyzing biomedical and genetic data across multiple institutions, enabling the use of AI tools in research. In essence, effective AI learning and applications require access to large-scale data, which this approach is designed to support. Much of today’s health data is stored in separate systems that cannot easily communicate with one another. FEAST helps bridge these gaps by enabling researchers to analyze distributed datasets while maintaining strict patient privacy protections.
By connecting information from different hospitals and research centers, scientists can better understand how cells respond to disease and treatment—knowledge that could help physicians identify which therapies are most likely to benefit each patient.
Studying Cancer One Cell at a Time
Cancer is rarely uniform. Even within a single tumor, individual cells can behave differently.
To capture these differences, Horvath’s team studies genetic activity at the single-cell level. AI-powered tools allow them to detect subtle variations in gene expression and identify molecular signals that might otherwise remain hidden.
The research group has also developed accessible computational tools that enable scientists worldwide to apply these AI-driven approaches to their own datasets, expanding the reach and impact of the work.
Why This Work Matters
Biomedical research now produces vast quantities of genetic and clinical data. Turning that information into meaningful discoveries requires new computational approaches.
Projects like FEAST allow researchers to analyze data across institutions and populations, revealing patterns that smaller studies cannot detect. These insights could accelerate the development of:
- Earlier and more accurate cancer diagnostics
- More precise treatment strategies
- Improved clinical trial design
Together, these advances bring medicine closer to a future in which cancer care is guided by each patient's unique biology.
About the Researchers
Anelia Horvath, PhD, MSc
Research Professor, George Washington University
Director, McCormick Bioinformatics Core
Director, MS program in Bioinformatics and Molecular Biochemistry, GWU
Dr. Horvath's research decodes tumor evolution at single-cell resolution by integrating long-read transcriptomics with AI to resolve cell-level mutational architectures, isoform dynamics, and allele-specific regulatory states. These frameworks convert high-dimensional signals into predictive, isoform-aware models that anticipate clonal trajectories and therapeutic resistance with unprecedented cellular precision.
Raja Mazumder, PhD, MS, MSc
Professor and Co-Director, McCormick Genomic and Proteomic Center
Dr. Mazumder creates biomedical databases and computational tools that assist researchers in understanding genetic information and applying it to disease prevention, diagnosis, and treatment.
