Children’s Hospital of Philadelphia Optimizes Pediatric Patient Research Using AWS HealthOmics

CHOP was the first US hospital dedicated to pediatric healthcare, and the CHOP Research Institute has advanced breakthrough pediatric treatments for rare diseases. “Getting a comprehensive view of our patients is crucial to delivering the best possible care, based on the most innovative research. Combining multiple data modalities is foundational to achieving this goal and discovering crucial links for treating them,” says Dr. Ingo Helbig, attending neurologist and scientific director of Project Omics at CHOP. “The challenge lies in making multiomic data jointly accessible with other datasets, and then using technologies like artificial intelligence and machine learning for better diagnosis and treatment.”

In 2017, CHOP launched Arcus, a suite of tools and services for CHOP’s research community that links biological, clinical, research, and environmental data. The CHOP Research Institute also built Arcus Omics, a growing library that is currently composed of 12,000 exome-genome datasets. This library set the stage for CHOP’s omics and big data strategy to drive personalized and precision medicine for children by linking the hospital’s clinical and research data to make it findable, accessible, interoperable, and reusable. “Our idea was not only to create an institutional gold standard data repository for omics data but also to promote its generation and usage for diagnostics and therapeutics,” says Jeff Pennington, chief research informatics officer and associate vice president at CHOP.

To facilitate this, CHOP had to remove data silos, not only within its own institution but also between healthcare organizations globally. CHOP also wanted to adopt technology that could scale as it continues to gather tens of thousands of patients’ exome and genome information. “While there was open-source software available for omics data analysis, it required a lot of engineering time and effort,” says Pennington. “The more we needed to scale, the harder it became.”

“AWS HealthOmics was an answer to my wishes for 20 years—we can now expand our understanding of our patients’ health, all the way down to their DNA,” says Dr. Helbig. “The managed service reduces the undifferentiated heavy lifting so that our bioinformatics engineers can work on issues close to child health rather than figuring out how to scale.”

CHOP uses AWS HealthOmics to analyze and generate insights from multiomic data—giving researchers more time for unlocking scientific discoveries to improve pediatric treatment and cure diseases—using joint variant calling on omics in addition to variant calling. “Our data volume quickly goes from terabytes to petabytes,” says Pennington. “AWS HealthOmics supports large-scale analysis and collaborative research without us needing to worry about provisioning the underlying infrastructure, and it avoids the time sink of engineering, configuration, and systems integration.”

The institute has created a model of digital data archiving that does not push everything into a standardized database. To meet guidelines from the National Institutes of Health on federally funded data sharing, primary and secondary data is packaged into collections by digital archivists engaging with data scientists, metadata librarians, and researchers. Outside of AWS HealthOmics, the team uses a range of AWS services for linking genomic data to clinical and other multiomic data.. “Using AWS, we not only find economies of scale for analysis and storage but also maximize the scientific usage of omics data,” says Pennington.

To address barriers around data accessibility, CHOP has developed a three-tier approach that makes Arcus Omics available to any researcher. The first tier of access is geared toward people who need fast answers to urgent clinical questions but do not have a data science background. The second tier offers a lab for virtual experiments. The final tier supports bioinformatics at the ground level. “A clinical fellow at CHOP will spend at least 2 years on research,” says Pennington. “They don’t have time or money to generate their own data. Being able to access CHOP’s datasets and do productive work during their fellowship could lead to funding that can enable them to continue their scholarly research.”

The platform helps researchers get access to searchable datasets in minutes rather than days. “Using AWS HealthOmics, we can have all this data in one database and query it with just a click, saving hours, if not months, of work in finding genes,” says Helbig. In one case, CHOP researchers discovered a genetic mutation in epilepsy patients by cross-referencing genetic epilepsy data with gene expression data collected from samples of brain tissue taken from surgery patients. “Vertical integration between different datasets and modalities helped us solve a decade-old mystery,” says Helbig.

Given the sensitive nature of its data, CHOP protects patient privacy with HIPAA-eligible AWS services, built-in data access, strict security controls, and an AWS HIPAA Business Associate Agreement. “AWS has been a great choice for setting up a cloud-based environment that addresses the challenges around privacy, security, data ownership and stewardship, governance, and fairness,” says Helbig. For patients, this translates to faster diagnoses, better treatments, and improved outcomes.