Hi All,
This is a fantastic use case, where semantically formal annotations
will be absolutely required in order to resolve even a portion of the
questions listed, if they are to cast a wide net across all the
relevant information - e.g., primary data repositories (individual,
as well as data warehouses and federated data resources), the
literature, clinical trial reports, clinical records, etc.
This is exactly what I was trying to get at back in the late summer
on the BIOONT Wiki. The idea was to go into the literature to pull
together several relevant papers collectively providing findings
related to a broader question in neurodegenerative illness. We'd
then go through the papers - just at a high-level - and pick out the
essential findings, expressing them using the PATO formalism http://
www.bioontology.org/wiki/index.php/PATO:Main_Page). This requires
expressing the observation made in the context of ontologically
defined entities from the molecular on up through behavior. In
performing this task, one must determine which of the available
community ontologies can cover the required domains. It also helps
to provide a clear categorical description of the type of data
repositories that must be integrated and where they fit into the
overall semantic landscape. BioRDF would then define what exists for
those domains that is publicly accessible - as is outlined somewhat
in the BMC Bioinformatics manuscript.
I apologize for missing the meeting today. I was ill and am slowly
coming back.
Re: the BIRN Use Cases relevant to the HCLSIG Demo:
As I mentioned in the TCon last Thursday, I've been working with the
BIRN Ontology Task Force to assemble a public page providing the many
use cases assembled by various BIRN research projects - all of which
focus on drawing together information from disparate sources to
address scientific questions related to Alzheimer's Disease,
Schizophrenia, Parkinsonism, and MS (with Huntington's Disease and
Autism being introduced over the past year or so). Given the heavy
neuroimaging focus on BIRN, many of the details of these use cases
relate to seeking correlations between segmented geometries in
various MRI, MRI/DTI, fMRI, light-microscopic, and EM images and
other related data extending from genotype & gene expression on
through behavioral & cognitive assessments. Many of the active
research projects in the BIRN involve the generation and analysis of
neuroimaging data and associated experimental assessments - data that
may be acquired in one lab, processed in another, and ultimately used
for analysis by many. Already there are 100s of TB of data on the
BIRN infrastructure, though imaging researchers will recognize this
is quite small, given that includes raw image data, and data in
various stages of processing and analysis.
BIRN's goal is to provide a high-capacity, high-throughput
infrastructure to support this manner of collaborative research and
to host the results for public use. The components assembled to date
consist of the following:
1) a distributed query mediator posting queries across the 60+ site
source data repositories - many of which are RDBMS-based. A source
repository needs to "register" with the mediator in order to be
accessible during query resolution
2) an ontology (BIRNLex) to provide a shared, formal expression of
the required biomedical entities and processes that will support
machine-parsing of semantic assertions defined on data via ontology-
based annotation; note that the BIRN ontology draws heavily on the
community ontology efforts throughput biomedicine, to avoid
duplicating effort and ensuring maximal semantic integratability
3) a storage grid (based on SRB) that includes a metadata catalog
which can be linked to the mediator; when a query result tuple points
to a binary object such as a 2D, 3D, or 4D image set stored on SRB,
this can be returned to the querying application - whether its an
interactive, end-user oriented query front end, or an automated
process, such as a distributed image processing pipeline
4) an XML Schema (XCEDE) and associated toolset for uniform
exchange of data. This can help to simplify and standardize the way
in which the 3 resources above are put into practice, as well as
provide a uniform view of BIRN-hosted information to the greater public.
As I mentioned before, the BIRN project as a whole has focussed on
providing this required infrastructure, tools, and data standards for
the current BIRN participants. Though the goal is for all of these
resources to be publicly available, human resource limits have meant
what is currently available via the BIRN infrastructure to the public
is very limited. There is an effort underway to address this issue.
We on the Ontology Task Force have made our current draft ontology
available publicly. What is being assembled in that ontology is
designed to support the use cases identified by the participating
research labs. These are grouped under "testbed" categories - two
human focused - Morphometry BIRN (MRI & DTI-MRI) and Function BIRN
(fMRI) - and one focussed mouse models of neurodegenerative disease -
Mouse BIRN. More recently, there have been efforts to include
neuroimaging projects associated with the various North American
primate centers, which brings in macaque, a species used extensively
for various structure/function studies in the brain (see
www.cocomac.org).
As a preliminary to assembling a comprehensive and coherent listing
of the use cases from throughout BIRN, I've assembled pointers to
"some" of the use cases on the public BIRNLex page:
http://xwiki.nbirn.net:8080/xwiki/bin/view/+BIRN-OTF-Public/Home
I would say the following are all worth looking through:
1) The recent Morphometry BIRN use case listed there on that page
provides a since of the comprehensive nature of the questions being
addressed in that project - in this particular case various ways of
determining whether diagnosed depression can effect the onset and
severity of AD.
2) The link to the Function BIRN Use Cases below the table on that
page provides a sense of the complex imaging & image associated
metadata one must semantically represented in order to make effective
use of fMRI data to answer a broad range of questions questions
related to the structure & function (or malfunction) of the nervous
system.
3) The "3 BASIC LEVELS OF DATA MEDIATION" link, since it was
assembled by the group in Mouse BIRN that works on the alpha-
synuclein mouse model of PD.
There are many, many more use cases than these, but it will probably
be a while before I'm able to do better than this. BIRN is large
organization, and I'd want to make certain along with the Use Cases,
we provide information on which are actually being pursued as
research questions by BIRN-associated projects, so as to be able to
indicate there is data related to that use case accumulating on the
BIRN infrastructure, and the BIRN ontology has covered the relevant
knowledge domains.
I hope this is helpful.
I will try to find time soon to get to more of the BIRN use cases,
especially those supported by data and ontological graphs and
directly related to the current PD use cases Don, Kei and others have
included in the BMC Bioinformatics manuscript.
Cheers,
Bill
On Dec 11, 2006, at 11:03 AM, June Kinoshita wrote:
Hi Susie,
Elizabeth, Gwen and I have been working on a fairly elaborate use
case. To help the con call participants follow it, we've prepared
this simple outline:
A hot topic in AD therapy is immunization - developing a vaccine or
antibody that will neutralize amyloid-beta peptide (Abeta), the
putative toxic agent that is hypothesized to cause Alzheimer disease.
Problem: The first clinical trial was halted because patients
developed encephalitis.
Question 1: Why did some patients but not all get encephalitis?
Question 2: There is debate about whether Abeta is the toxic
entity; but assuming it is, what form is toxic? Abeta has been
reported as monomers, dimers, soluble oligomers, protofibrils,
insoluble fibrils.
Research Challenge: If immunization therapy is to work, i.e. it
clears Abeta from the brain and does not cause encephalitis, one
needs to 1) identify the exact toxic form, 2) understand cause of
encephalitis (brain inflammation)
In our use case, an investigator reads about the discovery of a new
form of Abeta, called Abeta*56, that is reported to cause memory
impairment in a mouse model of AD.
Is Abeta*56 a good target for immunization therapy?
Question: Is there human data to support that Abeta*56 is involved.
A query of PubMed finds a paper reporting that a form of Abeta with
identical molecular weight, called ADDL, is elevated by as much as
70-fold in human AD patients' cerebrospinal fluid. A hypothesis
about ADDL causing memory loss in AD is posted on Alzforum.
Question: By what mechanism might Abeta*56 cause memory loss?
The ADDL Hypothesis on Alzforum suggests that ADDL (= Abeta*56?)
disrupts LTP.
Question: What is the mechanism of LTP, in a part of the brain that
is relevant to AD?
The literature indicates CA1 hippocampal neurons, and A- and D-type
K channels are involved in LTP. BrainPharm data state that CA1
hippocampal neurons have A-channels. What's more, the A-current is
reduced by Abeta.
Question: Would an antibody directed against ADDL / Abeta*56
restore A-current in the mouse model hippocampal neuron (e.g. in an
organotypic slice prep)?
A query locates an antibody to ADDL and where to obtain it.
Question 2 is what determines vulnerability to encephalilitis among
immunized patients?
A literature search finds interferon-gamma (INFG) may be a player.
This comes from both mouse and human trial data (need to check this).
Question: Do polymorphisms in INFG, or differences in INFG
regulation, correlate with inflammatory response to vaccine?
Question: Why did the mouse models not develop encephalitis in
preclinical studies?
Our investigator queries pathway databases to identify the gene
network involved in IFNG regulation, and also SNP databases for
differences between mouse strains, mouse and human. He narrows down
a group of genes and queries the AlzGene database to see if any
gene association studies have shown a correlation between any of
these genes and AD risk.
Our investigator proposes a study to genotype participants in the
failed vaccination trial to find out whether the encephalitis
response correlates with specific SNPs in the candidate genes.
On Dec 8, 2006, at 3:16 PM, Susie Stephens wrote:
Here's a reminder for Monday's BioRDF call.
Date of Call: Monday December 11, 2006
Time of Call: 11:00am Eastern Time
Dial-In #: +1.617.761.6200 (Cambridge, MA)
Participant Access Code: 246733 ("BIORDF")
IRC Channel: irc.w3.org port 6665 channel #BioRDF
Duration: ~1 hour
Agenda
Matthias Samwald and Alan Ruttenberg will be providing task updates.
Bill Bug, Elizabeth Wu and Scott Marshall will be discussing the
scientific queries that they have been researching.
Kind regards,
Susie
Bill Bug
Senior Research Analyst/Ontological Engineer
Laboratory for Bioimaging & Anatomical Informatics
www.neuroterrain.org
Department of Neurobiology & Anatomy
Drexel University College of Medicine
2900 Queen Lane
Philadelphia, PA 19129
215 991 8430 (ph)
610 457 0443 (mobile)
215 843 9367 (fax)
Please Note: I now have a new email - [EMAIL PROTECTED]
