Coming to grips with biological complexity at the NIH

1
Coming to grips with biological complexity at the
NIH

• Eric Jakobsson
• Director, NIGMS Center for Bioinformatics and
  Computational Biology
• Chair, NIH Biomedical Information Science and
  Technology Initiative
• For the Networks and Complex Systems Seminar
  Series
• Indiana University
• March 28, 2005

2
What are the elements of biological complexity?

• At the molecular level, combinatorial
  explosioneffectively an infinite number of
  possible gene sequences and protein structures.
  (Albeit a constrained infinite number.)
• At the reaction network level, the fact that
  single gene products are implicated in multiple
  reaction networks.

3
More generally (from Tong et al, Global mapping
of the yeast genetic interaction network,
Science, 2-6-04)
4
What is forcing us to come to grips theoretically
with biological complexity?

• On the technology-push sidecapabilities for
  high-throughput data gathering that have made us
  aware that biological networks have many more
  components than we previously surmised.
• On the biology pull side---the realization that
  to the extent that we dont characterize
  biological systems quantitatively in their full
  complexity, the scope and accuracy of our
  understanding of those systems will be
  compromised. (In classical experimental terms,
  the uncontrolled variables in the system will
  undermine our confidence in the conclusions we
  draw from the experiment or the observation.

5
How has coming to grips with complexity made
computation become critically important to
advances in biology?

• One cant organize the high-throughput data into
  structures that are useful for generating
  knowledge without computers.
• One cant do the analysis, modeling, and
  visualization without computers.

6
Top Ten Advances In Biomedical Computing In The
Last Decade

• 1. Sequence alignment tools.
• 2. Enabling Systems Biomedicine.
• 3. Identification of genes for disease
  susceptibility.
• 4. Computational model of HIV infection---1996
• 5. Tomography
• 6. Computer-aided Prosthetic Design.
• 7. Modeling electrical behavior of neurons and
  other electrically excitable tissue.
• 8. Dissemination of bioinformatics tools on the
  Web -- 1993
• 9. Telemedicine.
• 10. Establishment of computer networks for
  surveillance of disease.

7
Systems Biology is an approach to understanding
biological complexity. What are the elements of
systems biology?

• Make a parts list (based on experiment)
• Make a network diagram to see how the parts are
  interconnected. (Inference from experiment and
  theory)
• If you are able to do so, make a
  mathematical/computational model of the system
  that will test the completeness and correctness
  of your understanding and provide a guide for
  future experiments.

8
Differences between systems biology and
traditional cell and molecular biology

• Experimental techniques in systems biology are
  high throughput.
• Intensive computation is involved from the start
  in systems biology, in order to organize the data
  into usable computable databases.
• Exploration in traditional physiology proceeds by
  successive cycles of hypothesis creation and
  hypothesis testing data stores accumulate during
  these cycles.
• Systems biology initially gathers data without
  prior hypothesis creation hypothesis creation
  and testing comes during post-experiment data
  analysis and modeling.

9
A simple computer experiment using prior data
stores that were pre-hypothetically assembled.

• Question Is the plot of Clan of the Cave Bear,
  involving interbreeding between Neanderthal and
  anatomically modern humans, biologically
  plausible?
• Relevant data in Genbank Mitochondrial DNA
  sequences from Neanderthals, modern humans, and
  other mammals.
• Hypothesis Degree of relatedness of DNA may be a
  marker for possibility of interbreeding.
• Computer experiment Compare comparable DNA
  sequences from Neanderthals, modern humans other
  mammals including species that interbreed and
  those that do not.

10
First, search Genbank for Neanderthal DNA
sequences
11
(No Transcript)
12
(No Transcript)
13
Then, use one of the Neanderthal sequences as a
BLAST probe to find its most closely related
human sequence.
14
(No Transcript)
15
And also use the Neanderthal sequence as blast
probe to find corresponding sequences in other
mammals.
16
(No Transcript)
17
Next, align the sequences
18
(No Transcript)
19
And draw a phylogenetic tree
20
(No Transcript)
21
Elements of systems biology in the illustration

• Made a parts list from pre-existing
  comprehensive data
• Used a mathematical model to interpret functional
  relationships among the parts from the data
• Drew a tentative conclusion but need more data

22
What is the range of viewpoints on systems
biology?
23
Top ten challenges for computational biology in
the next decade

• In silico screening of drug compounds
• Predicting function from structure of complex
  molecules at an engineering level of precision
• Prediction of protein structure
• Accurate, efficient, and comprehensive dynamic
  models of the spread of infectious disease
• Intelligent systems for mining biomedical
  literature
• Complete annotation of the genomes of selected
  model organisms
• Improved computerization of the health-care
  delivery system
• Making systems biology a reality by integrating
  appropriate computational tools
• Tuning biomedical computing software to computer
  hardware
• Promoting the use of computational biology tools
  in education

24
Some important problems with biomedical computing
tools are

• They are difficult to use.
• They are fragile.
• They lack interoperability of different
  components
• They suffer limitations on dissemination
• They often work in one program/one function mode
  as opposed to being part of an integrated
  computational environment.
• There are not sufficient personnel to meet the
  needs for creating better biological computing
  tools and user environments.

25
II. We want you if.

• You have ideas that can contribute to realizing
  the NIH vision.
• You have the skills and motivation to implement
  those ideas.
• The right vehicle for realizing your ideas is
  EITHER a large project or a small project.

26
Looking at the NIH for support for computational
projects I General Issues

• First Principle NIH is a mission-driven agency.
  We support basic science (lots of it) and
  technology and infrastructure development (on an
  increasing trend line), but it all must be
  justifiable by a payoff down the line in
  improving the health of the American people.
• Corollary Principle We understand that the
  payoff may not be immediate, so we support work
  where the payoff is a decade or more in the
  future. It is better to present a justification
  for a reasonable but long-term payoff than an
  unrealistic short-term payoff.

27
Looking at the NIH for support for Computational
Projects II Perspectives on the Role of
Computation in Biomedical Research and Health
Care Delivery

• We see that non-trivial computation is critical
  to every aspect of our mission, from the most
  basic research to the efficient and effective
  delivery of health care in all venues.
• We see the corollary Inefficiencies, gaps, and
  flaws in computation are limiting the pace and
  scope of all aspects of our mission.
• We have only gotten the message recently, so we
  are a work-in-progress with respect to
  implementing our understandings about computation
  in programs and practices.
• We need computer scientists, computational
  scientists, and information technologists to be
  partners with NIH in getting it right.

28
Looking at the NIH for support for computational
projects III Finding out what NIH actually funds

• CRISP data base (Google NIH CRISP provides
  keyword-searchable database of all NIH-funded
  projects from 1972-2004
• Comprehensive access to publications by NIH
  grantees provided by author-searchable Pubmed
  literature database (Google pubmed)

29
Looking at the NIH for support for computational
projects IV Building on your knowledge of what
we now do to what we might support you for doing

• First-stop (but not one stop) information
  source is the BISTI home page (Google NIH
  BISTI), button under Funding
• If you dont find a funding announcement that
  fits your ideas/capabilities, but you feel you
  have something to contribute, dont hesitate to
  send an unsolicited application. (Receipt dates
  February 1, June 1, and October 1 each year for
  new applications). Success rates for unsolicited
  applications are often as good as, in some cases
  better than, success rates for proposals
  submitted in response to specific funding
  announcements.
• Consult with an NIH Program Director at the
  concept development stage. This is easy if you
  are responding to a funding announcementthe
  right contact information is in the funding
  announcement. For an unsolicited application,
  you may need to browse through Web sites for many
  of the semi-autonomous 27 Institutes and Centers
  that comprise the NIH, as well as the NIH Roadmap
  site, that contains information on NIH-wide
  initiatives. But---NIH is a strongly
  interconnected community, so if you start calling
  program staff and the first person you call is
  not the right person, you will get good direction
  to the right person fairly quickly.

30
Looking at the NIH for support for computational
projects IV Building on your knowledge of what
we now do to what we might support you for doing
(continued)

• Research study sections as well as programs
  (Google NIH CSR), button under Study Section
  Information.
• On study section targeting, consult with Program
  Director and/or Scientific Review Administrator
  (Understand that program and review functions at
  NIH collaborate with each other but are
  independently accountable. This is different
  from NSF, where the same individuals are
  responsible for both creating program and
  overseeing review. With respect to NIH review
  issues, the AUTHORITATIVE information comes from
  the review side)
• FOLLOW THE RULES AND GUIDELINES! (Google NIH
  398 in addition to particular funding
  announcements.) That gives program and review
  staff more time to deal with your scientifically
  substantive concerns, because they wont have to
  work around emergent procedural issues.

31
Looking at the NIH for support for computational
projects IV Building on your knowledge of what
we now do to what we might support you for doing
(final)

• Develop an NIH grant journal club (or
  comparable structure) at your institution where
  colleagues read and critique each others NIH
  grant applications and progress reports in
  preparation.

32
THE BEGINNINGThank you for your attention and
your commitment to building the future