How text mining is changing the way we tackle chronic disease

Researchers are working with Pathway Studio to design innovative treatments using simulations of complex diseases

Dr. Gordon Broderick (right) with team members Hooman Sedghamiz and Dr. Matt Morris at the Center for Clinical Systems Biology at Rochester General Hospital. By studying our biological wiring and redirecting the body’s regulatory programs, they are working to develop disease modifying treatments to escape complex illness and achieve long-lasting remission.
Dr. Gordon Broderick (right) with team members Hooman Sedghamiz and Dr. Matt Morris at the Center for Clinical Systems Biology at Rochester General Hospital. By studying our biological wiring and redirecting the body’s regulatory programs, they are working to develop disease modifying treatments to escape complex illness and achieve long-lasting remission.

Imagine being able to mine 40,000 articles in a matter of minutes and use the information to develop a solution to transform the lives of millions of people. That’s what Dr. Gordon Broderick, Director of the Center for Clinical Systems Biology at Rochester General Hospital (RGH), is doing. Using Pathway Studio, he and his team are creating networks of connections that capture the tightly choreographed interactions between different biological molecules and cells that make up our anatomy.

By working side by side with doctors, the engineers can use the model biological circuits to give insights that could lead to better diagnoses and guide the design of effective treatments for complex medical conditions that defy conventional approaches. To move this work forward, the team brings together a wide variety of skill sets that cross traditional boundaries. Clinical specialists from domains ranging from psychology to pediatric infectious disease work with control theorists and highly skilled programmers – often from the gaming industry – to create large-scale simulations on the latest high-performance computing platforms. In this virtual biology environment, they collaborate with researchers around the world to tackle some of the most elusive and complex illnesses that affect the function of the hormone, immune and nervous systems.

With cross-disciplinary training in chemical engineering and computational biochemistry, Dr. Gordon Broderick leads the Center for Clinical Systems Biology, a hospital-based transdisciplinary team of medical researchers focused on improving the standard of care in complex chronic disorders by examining the biology of information processing in health and illness.

A year ago, Dr. Broderick and his colleagues started working with Elsevier’s Pathway Studio team, led by Dr. Chris Cheadle, Director of Genomics Research for R&D Solutions at Elsevier, to build these complex biological maps. Pathway Studio, a software program that pulls associations between biological entities, such as cells and molecules, from millions of published papers, had a huge impact on their work, Dr. Broderick explained:

Working with the Pathway Studio team, in minutes we could mine close to 40,000 journal papers to connect 50 or more biological entities and produce a first comprehensive map of their regulatory interactions. Pathway Studio allows you to look immediately at how we collectively understand the data – very quickly you’ve got a survey across the whole research community of how we think the system connects. We can then simulate the behavior of the system dynamically with the methods we are refining; we’re essentially bringing these maps to life with simulations where we can test our understanding and evaluate solutions in silico.

By working closely together, the two teams are sharing their expertise. The Pathway Studio team mines the material for information, and Dr. Broderick’s team provides the coding for the simulations. The result is a range of dynamic models based on published research, covering in vitro experiments, animal studies and human trials, that reveal possibilities for better combination therapies for complex illnesses.

An engineer’s eye view

When Dr. Broderick first joined Rochester General Hospital, he asked that the lab be located inside the hospital. This was an unusual request but one that made sense for the group: sitting at the intersection of the computational sciences, clinical sciences and basic life sciences, the team needed daily contact with doctors who were working directly with patients and who understood the complexity of these diseases and the challenges faced in a clinical setting to improving the standard of care and the delivery of more effective treatments.

With graduate training in control theory, Hooman Sedghamiz leads the team's efforts in regulatory biology with an emphasis on algorithm development — deciphering the dynamics of illness and the escape trajectories towards health. He is one of five team leaders covering synergistic areas of research that include knowledge acquisition and computing (Mark Rice), network biology (Saurabh Vashishtha), behavioral systems (Tory Toole) and biological modeling (Matt Morris).

It’s this complexity that might provide the solution – and data engineers may hold the key. Dr. Broderick and the team look at the human body as a computer that executes programs. They’re using techniques rooted in computer science and applying them to biology, approaching chronic illness by working out how the body can become trapped in a sort of control program that perpetuates dysfunction. He explained:

Like computers, the programs our body’s systems run help us navigate the hazards we come across and help us return to a balance once the situation has resolved itself. Sometimes the challenges are so intense that the body has to call on another regulatory program. If we forget to turn these off, the result is a regulatory trap – a situation in which the body is too far away from home, so it sets up camp in its new state.

Chronic medical conditions like Gulf War illness (GWI) may occupy such regulatory traps and hence don’t respond to single treatments, and Dr. Broderick believes this is why: thousands upon thousands of interacting molecules are involved the body’s biological pathways, building up a complex web or network of systems and processes. When the body switches into a new regulatory program, the whole web functions differently – there is no one molecule or pathway that can be fixed independently of the others.

In these chronic or slowly progressing autoimmune conditions, there is a regulatory imbalance between parts of the immune system, and in its interactions with central nervous and endocrine systems. Dr. Broderick and his team believe that when it is severely challenged, the immune system may be forced to take extreme measures and engage failsafe response programs. Even after the threat has subsided, the immune system can remain locked into these failsafe programs, causing chronic conditions. Dr. Broderick shared his thoughts on GWI:

Imagine harsh chemically-laden environments encountered by ground forces during the 1991 Operation Desert Storm in the Persian Gulf. At least one in four servicemen and women returned from this deployment with a complex and debilitating constellation of neurological and musculoskeletal symptoms, including chemical hypersensitivity. It is now believed that these exposures, specifically to trace amounts of neurotoxins, combined with the stress from the constant threat of a combat theater, may have trained the brain’s immune defenses to maintain a chronically elevated level of vigilance, causing it to dramatically over-react to the slightest perceived threat, such as commonplace bacteria or household chemicals.

Supported by the US Department of Defense’s Congressionally Directed Medical Research Program (CDMRP), Dr. Broderick’s team has been working closely with neuro-toxicologists at the CDC’s National Institute for Occupational Safety and Health (NIOSH) to unravel the complexity of GWI.

His group focuses on determining how differently our cellular and molecular networks might be used in health and illnesses like GWI. Information flow through these networks provides detailed fingerprints capable not only of recognizing illness but, more importantly, pinpointing the possible processes involved. They are bringing these models to life through large-scale computer simulations that mimic the flow of biological information through the brain’s immune system. He explained:

Our approach focuses on effective intervention. In autoimmune illnesses like GWI this involves testing millions of strategies for teaching the immune system to stand down. This approach can be broadly applied to a myriad of conditions, in each case finding out how the body’s control programs switched to enforce new rules of play, with the aim of working out how to switch them back.

Mining, mapping, modeling

A cellular and molecular immunologist, Dr. Matt Morris is uniquely skilled to advance the team's design of high-fidelity models that represent the characteristic functional  features of immune and endocrine biology.

Dr. Broderick’s team works closely with other research teams to combine sophisticated computer models with experimental and clinical data. The first real proof of this multi-team cross-disciplinary approach combining molecular medicine with computational hutzpah came in the form of successful animal trials with the CDC/NIOSH, providing early and encouraging evidence validating a scripted course of combination therapy predicted using large-scale computer models of Gulf War Illness (GWI).

That is mind-blowing to me: in one fell swoop, we added a lot more credibility to the idea of these regulatory traps that might participate in the persistence of your illness. To see that that might, at least in part, be true was very exciting for us. Imagine if we could be even partially successful in human trials – what a success story that would be for systems biology.

It would be a success story for patients around the world, too. Dr. Broderick believes treatment needs to be more multifactorial and delivered in a premeditated, scheduled or time-sensitive way – a better directed treatment plan that will help us go beyond managing symptoms and start thinking about sustained remission.

And the approach is not tied to a specific illness or even a level of biology, such as a hormonal pathway. If it’s possible to say “I think this will cause this”; it’s possible to test this statement using a logic model. For example, Dr. Broderick is currently working with the MD Anderson Cancer Center on biobehavior modeling, looking at the flow of information linking psychological factors – like trust, worry and medical literacy – that dictate whether a patient will adhere to treatment. The more data they have, the better they can validate the model – and this goes for anything from cancer to depression. As Dr. Broderick noted:

Illness is dynamic, not static; in order to treat a complex illness, you need to understand how the body is currently fighting it, so you know how and when to intervene. We’re interested in resetting regulatory programs so that the body can treat itself – re-establish the equilibrium and let it remember how to regulate itself normally.

Enticing top programmers

One key to this success is the close collaboration Dr. Broderick and his team has with clinicians and patients. After considering going into medicine himself, Dr. Broderick instead studied mechanical engineering and chemical engineering and got a PhD in chemical engineering and applied statistics before completing a first postdoc in cancer genomics and a subsequent postdoc in computational biochemistry.

Taking that engineering experience back into the clinic gave him a different perspective on the data. At the time, early broad spectrum gene arrays were hitting the market and could measure 12,000 gene products at a time – a big step up from polymerase chain reaction (PCR) techniques that could show 40-80 gene products at a time. But it was still nowhere near what Dr. Broderick was used to, as he recalled:

I have to smile when medicine says it has big data. Banking has big data. Chemical refineries have big data. Medicine has broad data, but the datasets are shallow and there aren’t many samples. Still, people were losing their minds. I remember sitting in a clinical meeting, where we were going down this endless spreadsheet of the individual gene products. … it was mind-numbing. You could see the mindset was really struggling with the high dimensionality of it.

Experience from the chemical industry, where one could typically have 60,000 variables measured at tenth of a second intervals, provided Dr. Broderick an interesting new perspective on how systems biology might be introduced into a clinical setting. The discovery of the basic underpinnings of biology, though admittedly crucial, is not the primary objective of the Center – Dr. Broderick’s group is focused on translating these fundamental principles into protocols and tools that directly impact clinical care. This objective is inspiring the next generation of programmers – and enticing them away from some of the industry’s top tech positions. In Dr. Broderick explained:

We’re stealing programmers from the gaming industry, because we’re using the same coding. They can do the same really cool programming they’re doing for Google, but instead use it to try to understand Parkinson’s. We put them front and center, so they meet the very people they’re trying to help – they have to cross a hospital and a waiting room full of patients, who appreciate the work they’re doing. And with all the tech giants getting involved in health research, it’s a really exciting place and time to be doing this work.

Pathway Studio

By integrating a vast knowledgebase of biological relationships harvested from millions of published articles (full-text and abstracts) with analytical and visualization tools, Pathway Studio enables scientists to explore molecular interactions along with the cause and effect relationships associated with biological processes across multiple organisms. By harnessing the extensive content found beyond the public domain, novel discovery is both more likely and more fully supported. This integrative and comprehensive biology-based investigational framework is ideal for drug target identification, disease modeling, biomarker identification and drug repurposing. Pathway Studio draws from:

  • 10,000+ journals
  • 25 million+ PubMed abstracts
  • 200,000+ clinical trials
  • 4 million+ full text articles from 1700 journals
  • 6.1 million+ high-quality relationships extracted
  • 1,800 professionally-curated pathways

This is the knowledge warehouse, and partners like the Centre for Clinical Systems Biology at Rochester General Hospital (RGH) are working to make this knowledge actionable in the hands of first-line health care professionals.


Written by

Lucy Goodchild van Hilten

Written by

Lucy Goodchild van Hilten

After a few accidents, Lucy Goodchild van Hilten discovered that she’s a much better writer than a scientist. Following an MSc in the History of Science, Medicine and Technology at Imperial College London, she became Assistant Editor of Microbiology Today. A stint in the press office at Imperial saw her stories on the front pages, and she moved to Amsterdam to work at Elsevier as Senior Marketing Communications Manager for Life Sciences. She’s now a freelance writer at Tell Lucy. Tweet her @LucyGoodchild.


comments powered by Disqus