Industry news that matters to you.  Learn more

2011 ISB Systems Biology and P4 Medicine Symposium, Day 2

Reading time: 13 – 22 minutes

The first Institute for Systems Biology (ISB) symposium was held in 2002. Now in its 10th year, the Systems Biology and P4 Medicine conference provides a setting for some of the world’s most influential researchers who are leading the way in applying systems biology and systems theory to medicine and health care delivery.

Session 3: Systems Approach to Disease

The conference resumed today with a session focused on systems approach to disease and chaired by ISB co-founder and president Lee Hood, MD, PhD.

The first talk of the day was from Aimée Dudley, PhD, an Assistant Professor at the Institute for Systems Biology. She described some of her work doing systems genetics in yeast.

  • Not many scientists get genotyped, with one exception: women in genetics labs preparing to have kids. (shows photos of her kids)
  • Use data-driven, network-based analysis to predict phenotypic outcome of a genetic or environmental perturbation.
  • Look at diversity on three scales: genotype, phenotype and environment.
  • Automation yields information beyond scale.
  • Use next generation linkage analysis to identify genomic regions linked to trait, predict causative polymorphisms.
  • It’s important to get deep phenotypic information. Shows data using 3D reconstruction to characterize yeast “fluffy” phenotype.
  • Developing image-based shape classification of yeast colonies.
  • Genetic regions predicted by Markov logic network can be used to identify many markers linked to trait.
  • Starting to characterize gene-environment effects.
  • Important for systems biology and systems genetics: if we really understand the system, we should be able to write accurate computational models.

Eric Schadt, PhD, Chief Scientific Officer from Pacific Biosciences, spoke next about a multi-scale biology approach to understanding and treating human disease. Before he began his talk, Lee Hood announced that Schadt was moving to Mount Sinai to head a new Institute of Genomics and Multiscale Biology.

  • Describes technology to watch DNA polymerase read DNA in realtime.
  • NEJM study, rapid identification of Haiti cholera outbreak strain.
  • Initiated project Nov 6-7, five strains sequenced in 1-1/2 days, published paper Dec 9.
  • At 12x coverage, able to resolve major structural features on Haitian isolates in 90 min.
  • Makes a case why we need structural accuracy: bacterial strains often differ in gene order and order is needed for good ID. Structure also needed to understand and predict virulence.
  • Only long reads allowed us to assemble long region with multiple copies of ribosomal genes.
  • Long reads across variant sequences allow you to avoid playing the informatics game of assembly.
  • Major structural features strongly point to south asia origins.
  • Used integrated startaegy, 36bp illumina + 454, long reads from PacBio, average span length 7000 kb spans.
  • In addition to structural variation, small nucleotide variations can be detected.
  • There are 15-16 DNA base modifications known, PacBio technology can resolve them based on polymerase kinetic shifts.
  • Identifying chemical modifications in DNA sequence is strand specific.
  • The ability to resolve epigenetic markers, identify markers and DNA damage will be very powerful in future genomics studies.

Joseph Nadeau, PhD, the Director of Research and Academic Affairs, and Professor at the Institute for Systems Biology, spoke about phenotypic variation and missing heritability.

  • Why do we study populations? Because that’s were the statistical power is.
  • Three stories: fractial genetics (diet-induced obesity), gene discovery in a fractal genome (obesity and glucose homeostatis) and transgenerational effects (looking at diet-induced obesity and feeding habits).
  • Individual genetics show many genes with strong effects that disappear in a population.
  • Systems properties may be more important than individual genetic effects.
  • Mouse data shows genes that increase weight, and genes that decrease. One variable can mask another. We need to have context.
  • We spend ~$150k/yr on high fat mouse chow. That’s not easy in mouse research.
  • Suggests epigenetics means genetic analysis should go beyond base differences in DNA.

A session panel discussion was then held where the audience could ask the speakers questions. Here are a few remarks:

  • Schadt: Modified bases are frequent phemonena in DNA; it opens up a new world.
  • Dudley: Networks are central to the idea of genetic analysis.
  • Nadeau: To me, it’s more about questions. The question is in the biology, disease is not only a dysfunction of particular genes, but also of a network.
  • Schadt: How do we perform network biology-based representation of biology at multiple scales
  • Schadt: That’s the beauty of genetics, you don’t have to understand deep molecular function to do it.

Session 4: Ethics, Policy and Economics

Mauricio Flores, JD, Managing Director of The P4 Medicine Institute chaired the fourth session on ethics, policy and economics. The panel discussion included Nancy Andrews, MD, PhD, Vice Chancellor for Academic Affairs and Dean of the Duke University School of Medicine; Sanders Williams, MD, President of the Gladstone Institutes; and Steve Gabbe, MD, Senior Vice President for Health Sciences at Ohio State University and Chief Executive Officer of The Ohio State University Medical Center. Below are some comments captured during the discussion:

  • Steve Gabbe: Transformation of healthcare is going to impact research, educational and patient care, and administrative processes.
  • Nancy Andrews: As a medical school dean, I think a lot about how the healthcare workforce is going to change.
  • Nancy Andrews: I think that our goal should be anticipatory health, a way to work towards keeping people healthy using many kinds of personalized information.
  • Sanders Williams: There’s two problems with US healthcare: a dysfunctional system and a lot of unmet medical needs.
  • Sanders Williams: I see hope in industry-academia partnerships to fix some of the healthcare problems.
  • Sanders William: We need to look to science for solutions to problems with healthcare.
  • Nancy Andrews: I have concerns about IP being a barrier to industry-academia partnerships.
  • Panel agrees: P4 medicine is more likely to be recognized in Europe where there’s a single payer.
  • Steve Gabbe: P4 medicine may happen first in Cuba, great preventative care, although poor for disease. Singapore, where care is logically organized.
  • Steve Gabbe: I feel it is essential that the physician-patient relationship be maintained in future medicine.
  • Steve Gabbe: In future medicine, physicians are going to need to be able manage teams of people and lots of data.
  • Nancy Andrews: I think that for future medicine we need to be very deliberate about matching medical jobs to the skills.
  • Sanders Williams: Compare medicine today and tomorrow’s needs to aviation industry: two physicians guiding hundreds of patients, low risk of death.
  • Steve Gabbe: Wellness begins with healthcare providers.
  • Steve Gabbe: Bringing P4 medicine and concept of prevention to medical school.
  • Nancy Andrews: Everybody in the room knows wellness, although many don’t necessarily follow it.
  • Sanders Williams: There’s a 40 billion market for alternative medicine, most of which is quackery. Why isn’t there a scientifically useful alternative?
  • Sanders Williams: People want wellness and are looking for shortcut to it. There is an opportunity.
  • Steve Gabbe: We need to address simple issues for people: safe places to exercise, fresh fruits and vegetables.
  • Sanders Williams: I don’t see a successful outcome to the US healthcare system; things are going to get worse before they get better.
  • Nancy Andrews: We should be investing in informatics, finding ways to handle and integrate huge amounts of information.

Session 5: Technology

The fifth session focused on technology and was chaired by Rob Moritz, Associate Professor and Director of Proteomics, Institute for Systems Biology.

First up was Gary Siuzdak from The Scripts Research Institute and Lawrence Berkeley National Laboratory. He spoke about mass spectrometry-based metabolomics as a unique biochemical approach for therapeutic discovery.

  • Why do mass spectrometry on therapeutic metabolomics? There’s an observable measure of phenotype.
  • Inhibitors used to be called anti-metabolites.
  • Our group is developing XCMS (open source freeware) to do comparative analyses.
  • metaXCMS, 2nd order meta-metabolomics analysis (Anal. Chem. 2011), reduced data down from22k to 3 candidates.
  • METLIN Metabolomics Database, >40k metabolites, >4500 metabolites with MS/MS data.
  • We’ve now combined all tools and it will be available online in the next month.
  • We’re using metabolomics in a mouse TNT model to identify new pathways as targets in chronic pain.
  • Chronic pain is a practical application. After a wound heals, pain can persist. We looked for metabolite changes where nerves attach.
  • We found DMS upregulated in the dorsal horn; it induces pain in rats.
  • We observed that NOE (an anti-metabolite) inhibits pain after injury.
  • We looked at the aging process in worms from a metabolic perspective; metaXCMS analysis reduced 150k features to 7, 4 of which overlap in human.
  • Spatially defining disease using nanostructure-initiator mass spec (NIMS).
  • We’re now trying to develop our informatics platform even further to understand fundamental biochemistry and anti-metabolites.

Jim Heath, PhD, from the California Institute of Technology spoke next about TCR engineering as adoptive immunotherapy for melanoma.

  • For adoptive T cell immunotherapy, the immune system is engineered to recognize and attack tumors. It gives patients immunity against their own cancer.
  • Current clinical trial has looked at the response of 15 patients to adoptive T cell immunotherapy, most relapse in 6 months.
  • We’re using nanotech and microchip immune monitoring technologies to assay individual cells.
  • We performed 1 million quantitative protein assays from single cells (numbers at the level of genomics!).
  • I believe we actually have a therapy that can respond faster than cancer can evade and knock tumor levels down to baseline.
  • Thermodynamic analysis can simplify network changes.

George Church, PhD, spoke about technologies for collecting and integrating genome, environment and trait data.

  • There’s some great genome and trait datasets out there, but some aren’t public, some don’t have traits, etc…
  • Genomes + Environments = Traits PersonalGenomes.org
  • There are 16k volunteers in the Personal Genome Project database.
  • The cost per genome is coming down to $4k genomes (in bulk for Illumina, Life Technologies).
  • What’s the future of next gen sequencing? Full chromosome haplotypes, zero gaps.
  • We need open access software. Gene sequencing toolmakers can distinguish themselves in other ways.
  • We used in situ sequencing to detect rare cells resistant to cancer drugs.
  • CDR3 in antibodies, 0-140 bases with sequences that don’t exist in the genome.
  • Every parent who sees a kid with a developmental anomaly will want genome sequencing.
  • We’ve looked at 22 hiPS lines and have found 1-12 coding changes.
  • People working with induced pluripotent stem cells should all be doing exome sequencing.
  • We are recruiting families to get sequenced. It helps to narrow down variables, but it’s not sufficient. We need new technology.

A session panel discussion was then held where the audience could ask the speakers questions. When asked about standardization, each speaker had a different answer:

  • Church: Kits are one way of maintaining standards, another way is a centralized service.
  • Siuzdak: The initial prep is more important than how we’re doing the analysis.
  • Heath: The ways we stimulate cells are well-accepted, the biggest frustration is patient samples.
  • Heath: You want to remove time and people for better standards.

Here are some speaker remarks to the question of the importance of open source and crowd sourcing:

  • Church: When setting standards, dealing with the FDA, and enhancing communication in a rapidly moving field, it’s nice to have open source – it’s not a great business model but it’s an important part of the ecosystem.
  • Siuzdak: XCMS is open source, meta-XCMS will be based at Scripts. It will test the willingness of users to upload data; it won’t be publicly accessible but it will be accessible to other investigators.
  • Heath: I’m a strong believer in open source, in addition to data, especially with new technologies, it makes it easy to get people using the tools.

The closing keynote address was given by Huntington Willard, PhD, the Nanaline H. Duke Professor of Genome Sciences and Director, Duke Institute for Genome Sciences & Policy. He spoke on the rising tide of personalized medicine.

  • Genomic medicine is the use of large-scale genomic information related to an individual’s genome, proteome, transcriptome, metabolome and/or epigenome in the practice of clinical medicine.
  • Genomic medicine is distinct from the specialties of medical genetics or genetic testing.
  • Genomic medicine uses large-scale data and creates analytical and cultural changes. Personalized medicine uses genomic data and is the choke point for P4 (or 5 or 6) medicine.
  • Early targets for genomic and personalized medicine: cancer, cardio health, infectious disease, type 2 diabetes, pharmacogenomics, toxic exposure, obesity.
  • There are several signatures in use in medical practice today (oncology and cardiovascular disease).
  • The infectious disease project at Duke has DARPA funding to find signatures for presymptomatic detection and diagnosis of illness resulting from infectious pathogens.
  • Duke viral challenge studies are working to predict who will develop upper respiratory viral illness.
  • College freshman, in dorms, are great incubators and models for monitoring infectious agent epidemics.
  • People claim to have discovered 150,000 biomarkers but only 100 have made it to the clinic. Why? Because we do not describe informatics for reproducibility.
  • In a field that is constantly transforming data, we need to have standards; reproducibility is critical.
  • The field needs to reevaluate how to effectively translate proteomic, genomic data.
  • We need to have series of standards to insure science is reproducible.
  • The problem for data handling: what is a “signature”?
  • Hand-written notebooks or transposition to computer are not a suitable form of storage for computational biology data.
  • We should all quit using Excel because we can’t find mistakes in data transformations; rows and columns change, cells contain variable data.
  • Quadra is a verifiable platform for genomic analysis. It contains QA-Tags, algorithmically and deterministically generated hashes for each piece of data.
  • Getting buy-in is hard, changing old practices to new methods and detail tracking is hard.
  • The unsequenced centromeres have been sequenced, plenty, just not assembled and annotated. They’re in the “unbin”. (picture of 24 network diagrams of the repeats look like abstract art)
  • There are characterized gene islands within centromeric repeats.
  • The cultural changes happening now are much greater than the changes that occurred during the molecular biology era.
  • Cultural and organizational models: university-based (Duke institute for Genomic Science and Policy), integrated health systems (Geisinger), stand-alone institutes (ISB, Sage Bionetworks), university-linked institutes (Broad) and the private sector (23andMe).
  • Challenges raised by data-intensive science and medicine:
    • capacity and design of institution computational infrastructure
    • culture of institutional and organizational units
    • culture of doing science – individual vs shared goals
    • system of academic rewards/models of career progression
    • culture of training and education
    • space design and assignments – wet and dry
    • faculty appointments – departments, schools
  • To be successful, universities and medical centers will need to be adaptive and look outward for best practices.
  • Educational mission:
    • blending of students in computational biology and genetics/genomics
    • “pre-grad” undergrads are open-minded, goal-oriented
    • postdocs have a wide range of backgrounds, diversifying
    • MD fellows are a potential group at risk. Why? Very few have time for the deep training necessary to compete.

With the conference coming to an end, Lee Hood made some closing remarks. “What are the biggest challenges to moving P4 medicine forward?” His answers:

  1. IT for healthcare is a gating factor for moving P4 medicine forward. Obama put $20 billion into the budget for healthcare IT. We’ll throw it away in 5 years. It won’t really address the big data integration problem.
  2. Education is a problem at every level: patients, students, ourselves, policy makers.
  3. The ongoing clash between big science and small science. Agencies fight big science. It would be disaster to kill it. Big science complements small science.
  4. We need families to analyze with everything we have. I despair about whether in the U.S. we can sequence enough people; China will. IRBs too much of an obstacle here.

More notes on each of the talks can be found on FriendFeed (ISB 10th International Symposium) and Healthcare Hashtags for Twitter (ISB2011P4).