hi kei,
yes, the proposal for specifying the dataset seemed to allow rich
annotation, one reason i liked it.
the piece i find missing is this. not being a biologist or statistician i
can't offer much help (my strength is software).
once a large group of individuals have been part of a GWAS (perhaps as part
of a clinical trial) and a marker based on a gene expression signature over
n genes has been determined, presumably that would be published as the set
of n genes and some averaged measure of up regulation or down regulation per
gene based on averaging that group of individuals and an outcome such as bad
prognosis or good prognosis associated with the marker. now if a new
individual has a gene expression profile, how will the expression of this
individual's genes be compared against the marker to determine which group
the individual falls into?
(there are other scenarios, such as multiple markers associated with
different outcomes but the above seems the simplest case.)
look forward to meeting you and the others at the F2F next week.
cheers,
michael
Michael Miller
[email protected]
----- Original Message -----
From: "Kei Cheung" <[email protected]>
To: "mdmiller" <[email protected]>
Cc: "Helen Parkinson" <[email protected]>; "HCLS"
<[email protected]>; "Tony Burdett" <[email protected]>
Sent: Thursday, October 29, 2009 1:43 PM
Subject: Re: Action Items from call today
Hi Michael et al,
Thanks for pointing to this generic approach of RDF representation of
datasets. A list of differentially expressed genes may be associated with
values such as P-values, fold-change, gene symbols, etc. Also, it's
important to capture metadata/provenance associated with the gene list.
This may include the type of statistical test (e.g., ANOVA) and the array
platform employed (e.g., Affymetrix U133A). This may be an interesting
discussion topic at the F2F meeting.
Cheers,
-Kei
mdmiller wrote:
hi all,
on a different HCLS thread i saw this proposal from jeni tennison for
specifying a generic dataset, it might be a useful way to encode a list
of differentially expressed genes. it looks like one could do this
encoding on the fly, so that the data itself at the source could be in
whatever format is natural.
http://sw.joanneum.at/scovo/schema.html
cheers,
michael
Michael Miller
[email protected]
----- Original Message ----- From: "Kei Cheung" <[email protected]>
To: "Helen Parkinson" <[email protected]>
Cc: "HCLS" <[email protected]>; "Tony Burdett"
<[email protected]>
Sent: Wednesday, October 14, 2009 7:03 PM
Subject: Re: Action Items from call today
Thanks, Helen.
To make it more concrete. I've been thinking about some example queries
that I hope can be answered by the RDF data once converted. I wonder if
the following example quereis can be answered:
Retrieve a list of differentially expressed genes between different
brain regions (e.g., hippocampus and entorhinal cortex) for normally
aged human subjects.
Retrieve a list of differentially expressed genes for the same brain
region of normal human subjects and AD patients.
Using these lists of genes one can issue (federated) queries to retrieve
addtional information about the genes for various types of analyses
(e.g., GO term enrichment).
Just a thought.
Cheers,
-Kei
Helen Parkinson wrote:
Hi
here are my action items from the call today
1. MAGE-TAB->RDF, Lena requested details.
Code here: https://sourceforge.net/projects/limpopo/
Java Parser for MAGE-TAB developed by EBI, used by several groups.
Contact Tony Burdett [email protected] for details. Tony estimates for
a simple RDF dump a few days work. Lena if you are interested in
working on this java code please contact Tony as he's already designed
with rdf export in mind
2. MAGE-TAB->MAGE-ML - code from Junmin Liu at UPenn
https://sourceforge.net/projects/tab2mage/files/ - see mage2tab
Pretty much all public MAGE-ML comes from AE and is available from
Arrayexpress ftp dirs as mage-tab already. Exceptions are Rosetta's
mage-ml importer, and non public data
3. EBI experimental factor ontology (EFO) slides, attached
see also www.ebi.ac.uk/efo
4. Noted that an RDF dump of atlas data and triple store access will be
useful, we'll announce when these are available
thanks
Helen
