Biology Reference
In-Depth Information
organ system can then be used to narrow down the
disease possibilities further. Once a disease is distin-
guished, yet finer resolution molecular signatures can
be used to help determine perturbed networks and
hence the best therapy options. 3) Considering the
plethora of computational challenges above, there are
many instances where it is hard to imagine amassing
sufficient statistical power to address all the relevant
states of wellness and disease. The key in these prev-
alent cases will be to leverage deep biology and
knowledge of mechanisms. In this case, the mapping
of molecular networks
sufficient sample numbers across a sufficient breadth of
the population to identify the most robust signatures.
Thus, there will be enormous commercial opportunities
that will form the basis for emerging health information
companies that will mine this data and produce content
that is directly usable by consumers (patients) as well as
physicians. While the number of signatures that are
translated currently is small relative to the number that
have been reported in papers, reasons for this from
a statistical point of view
given that we are deeply in
e
the small sample regime
are clear. There is every
reason to believe that as the data are integrated at the scale
that is necessary, with the rigor that is necessary, and with
the connection to biological networks and mechanisms
that is necessary, these approaches will indeed transform
the practice of medicine. And they will enable all of the
ideal features of diagnostics
e
through systems biology
approaches
and the interplay between genomics and
the environment
e
will be crucial to deciphering what
signals of the quantified self really matter to disease
treatment and health maintenance.
e
early detection, assess-
ment of the stage of disease progression, stratification of
disease, following the response to therapy and detecting
reoccurrences of disease.
Systems medicine provides powerful approaches for
dealing with signal-to-noise issues and biological
complexity. Systems medicine allows us to reduce enor-
mously the dimensionality of the search space for accurate
and robust biomarkers. For example, systems approaches
have led to organ-specific, cell type-specific and organ-
elle-specific biomarkers reflecting the functioning of key
disease-perturbed networks. Such approaches have also
been used to identify biomarkers in the blood, using
secreted proteins, proteins cleaved fromthemembranes of
the disease-perturbed cells, cytoplasmic and nuclear
proteins reflecting the death of cells, etc. The important
point going forward is to start with a narrow and targeted
set of biomarkers to search for molecular fingerprints that
can reflect the disease, including early diagnosis, stratifi-
cation of the disease types, and assessing the progression
of the disease. Accordingly much smaller populations of
patients can be used to identify valid biomarker signatures
with focused studies that leverage biological knowledge.
Exactly this approach has been applied successfully to
several mouse model diseases.
These five pillars of systems medicine together with
the digital revolution have given rise to the medical
opportunities embodied in P4 medicine
e
Health information systems of the future. Looking
forward, the development of P4 medicine based on
'omics' data will require extremely large repositories of
data in minable health information systems where
signatures are constantly evaluated, updated, locked
down and then re-evaluated for efficacy. Such systems
will be based largely on data in the 'real world' of
patient treatment and clinical outcomes
since this is
where the vast majority of medically relevant data will
come from in the future. Learning what factors most
affect patient outcomes by broadly measuring new data
sources and linking these back to activated patient
communities will serve as a powerful paradigm for
developing new tests to move forward iteratively
through clinical evaluation. Importantly, these systems
will be unbiased in the sense that they will record both
positive and negative outcomes as seen in the clinic
equally (a key problem with current literature practices,
where essentially only good outcomes are widely pub-
lished and reported in the development phases, and bad
outcomes are selected against). Such systems will need
to be very expansive in terms of numbers of samples and
measurements to be integrated
e
and there will thus be
institutional barriers to sharing data that will need to be
overcome through collaborative models. Whether the
data will be formed as raw data or processed into met-
adata for storage is a critical question, and almost
certainly we will move toward the storage of metadata to
reduce the data dimensionality. As an example, the 6
billion nucleotides of the human genome can be
compared against a reference genome
e
prediction,
e
prevention, personalization and participation.
and since
humans differ by about 0.1% of their genomes we could
store only the differences, reducing the data dimension-
ality by three orders of magnitude. There will be a strong
financial incentive to collaborate, as the clear winners in
this area will come only from those who can achieve
e
P4 MEDICINE
Systems medicine is focused on developing biological,
technical and computational tools to decipher the
complexities of disease. P4 medicine employs the strategies
and tools of systems medicine for quantifying wellness and
