Re: Mashup browser compatibility - and use of Mashup for Mouse BIRN contribution to future demo

On May 13, 2007, at 1:54 AM, Alan Ruttenberg wrote:

>
> On May 12, 2007, at 11:31 PM, William Bug wrote:
>
>> Works for me now (MacTel running OS X v10.4.9 & Firefox v2.0.0.3  
>> with a ridiculous number of plugins installed).
>>
>> This is great, Alan.
>>
>> So if I understand the changes you've made, we are now looking at  
>> a side-by-side comparison of the pseudo-colored, gene-specific  
>> immunohisto. sections from two different brains - (the brains  
>> 'gene painted' for Kcnip3 & Kcnd1) - presenting the roughly  
>> equivalent sagittal sections (Kcnip3@2850 and Kcnd1@2800) from  
>> those two brains?
>
> Yes. Just to warn - I need to rerun the Allen translation. I've  
> noticed at least two things 1) There are fewer genes that are in  
> the KB currently than I though I had loaded - don't know what  
> happened. And 2) I need to use the newest purl.org uri scheme and  
> do a mapping to entrez gene, as I realized that wasn't in the  
> initial mapping.  Also, I want to merge the ABA ontology with the  
> sciencecommons ontology, so there may be some changes in predicate  
> and class names to conform with the style we finally settled on.  
> I'll load this in to a new graph and then tell you when it is ready  
> and what has changed.
>
> I've also noticed the effects of the lack of spatial alignment  
> between images. You can see it clearly in another version of the  
> exhibit example. So high on my list when talking with them will be  
> to get some spatial information.
>
> The MBAT tools sounds cool. Do you think it should go into the sfn  
> abstract?
>
> -Alan

Thanks, Alan.  That's all useful detail to know.  I'm trying to make  
certain they hard code as little as possible regarding how the SPARQL  
queries are constructed and how the URLs are built from the SPARQL  
results, and this info will be very helpful in guiding those tasks.

Regarding mention of this MBAT work in the HCLS SfN abstract, I do  
make just a small mention of it at the end along with other data sets  
and ontologies we plan to bring into play.  I don't know how this  
would impact authorship.  Personally, I don't think it should effect  
authorship of the HCLS abstract now, since Mouse BIRN hasn't anything  
to show on this front yet.  The work so far has been getting the  
coders familiar with Jena and how to post the SPARQL queries and  
parse the results.  They've also done a review of the MBAT code to  
list the impact this work has on the current code base.  On the other  
hand, the MBAT abstract being worked up should definitely include  
you, if they are going to make explicit mention of using this  
mashup.  I don't know how that will play out given the extreme  
abstract length limits and other MBAT priorities.

In the end, a month prior the SfN meeting, they ask you to note other  
abstracts related through direct collaboration to the work reported  
in your abstract,   They use this when they create the abstract  
database for the 30K attendees.  The typical etiquette is to also  
make mention of the related abstracts in your poster; and/or slides,  
so we can certainly cross-ref each other, should this actually be a  
part of the MBAT presentation at SfN - which I'm doing everything I  
can to make happen.

Re: the degree of mis-alignment across ABA "equivalent" sections,  
remember their current registration (spatially aligning each  
individual brain to the underlying Franklin & Paxinos derived C57Bl/ 
6J brain atlas) has an error of between 100 - 300 microns with it  
mostly coming up at the higher end.

When trying to determine which section from brain X is "equivalent"  
to brain Y based on this atlas registration, the errors are roughly  
additive, which under the worse conditions will essentially be ~0.5  
mm.  The bounding box for an adult mouse brain is roughly:
	rostral-to-caudal axis (back to front - from olfactory cortex to the  
start of the spinal cord): ~ 20 mm
	dorsal-to-ventral axis (top to bottom - from the top of the parietal- 
frontal cortex boundary to the bottom of the hypothalamus): ~6 mm
	left-to-right or sagittal axis: ~10mm

So when the error gets up to 0.5 and your essentially trying to  
compare cell-level (~10 micron) visual patterns, things are bound to  
be off significantly.  I don't think this can be remedied much, until  
the ABA registration process resolution improves.  It IS likely to  
improve over the next year by at least a factor or 2 - 3, and since  
the brain to brain comparisons are essentially twice the registration  
error, we should se the brain to brain comparisons improve  
considerably when that happens.  That's all for ABA (and their image  
processing/alignment/registration collaborators) to work out.

Cheers,
Bill


>
>> Work has begun on the Mouse BIRN Atlasing Tool project to link  
>> this into our framework.  The idea would be, when viewing atlas- 
>> registered images or spatially-mapped gene expressions results  
>> overlaid onto an MBAT atlas view, a user would be able to right  
>> click on a an image and would be presented with a very simple way  
>> to query against your neurocommons ABA RDF store.  If they click  
>> on a spatially mapped gene-expression pattern, they'd be presented  
>> with a very simple option:
>> 	1) view Nissl & immuno-stain for the ABA brain stained for that  
>> gene - constrained to the set of sections from that ABA brain  
>> known to intersect with the selected brain region
>> 	2) view this side-by-side comparison by providing a user the  
>> ability to select from a list of other genes (preferably two lists  
>> - one constrained to genes we can relate to the current expressed  
>> gene pattern they just clicked on via www.genenetwork.org or via a  
>> pathways ontology; the other way to select a gene would be through  
>> a text query box - just give any string and we'd return all gene  
>> names that match via a regex either to the full gene name or gene  
>> name abbrev. tag)
>>
>> If they just click on an image with segmented brain regions (no  
>> gene expression), they'd provided with a way to again act on those  
>> two options above (Nissl + immuno for 1 ABA brain OR immuno  
>> comparison view for 2 brains) - selecting genes based on some  
>> combination of string matching and ontology constraints based on  
>> gene networks, pathways, genes related to a specific GO MF or GO CC.
>>
>> The results from the SPARQL query would then dynamically populate  
>> your Google Map HTML+JS template which would be presented in a  
>> JFrame (MBAT is a Java app).  Our developers are also spec'ing out  
>> how to add an additional "slicing" widget to the Google Map  
>> widgets, so a user could process through one of these constrained  
>> image sets (constrained based on their 'cutting' through a given  
>> brain region) from a given ABA gene-painted brain.
>>
>> I don't see any reason why most of this functionality wouldn't be  
>> ready to include in the SfN presentation (possibly even in alpha  
>> for ISMB in July, pending our access to MBAT programmer resources  
>> for whom this may not be the top priority).
>>
>> Cheers,
>> Bill
>>
>> On May 13, 2007, at 12:45 AM, Alan Ruttenberg wrote:
>>
>>>
>>> Hi Don,
>>>
>>> When you or someone else gets a chance, give a try again. I think  
>>> it will work now, thanks to:
>>>
>>> Brian Gilman (use a modern Ajax framework) and
>>> Mike Travers( wrap parentheses around the expression to be  
>>> evaluated).
>>>
>>> Much appreciated!!
>>>
>>> -Alan
>>>
>>> On May 12, 2007, at 8:26 AM, Donald Doherty, Ph.D. wrote:
>>>
>>>> Alan,
>>>>
>>>> I'm getting the same errors in Firefox 2.0 and IE 6.0 on the  
>>>> Windows XP
>>>> platform.
>>>>
>>>> Don
>>>>
>>>>>
>>>>> On the demo page there is a link to the Google Maps/Allen Brain
>>>>> Institute/SPARQL mashup. It works properly on Safari. However  
>>>>> it gets
>>>>> an error in Firefox after retrieving the results and doesn't  
>>>>> display
>>>>> an image. In IE7, it retrieves the results but then doesn't  
>>>>> seem to
>>>>> to update the images shown properly. The firefox test is on the  
>>>>> Mac,
>>>>> the IE7 test is under XP.
>>>>>
>>>>> I suspect that this has to do with details of the  
>>>>> XMLHTTPRequest. If
>>>>> there is someone knowledgeable about such issues I'd appreciate if
>>>>> you could have a look and see if you can figure out what the  
>>>>> problem
>>>>> is. Also testing on other browsers/OS would be appreciated.
>>>>>
>>>>> The URL to test is http://hcls1.csail.mit.edu:8890/map/
>>>>> #Kcnip3@2850,Kcnd1@2800
>>>>>
>>>>> Thanks!
>>>>>
>>>>> -Alan
>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>> -- 
>>>> Donald Doherty, Ph.D.
>>>> Brainstage Research, Inc.
>>>> www.brainstage.com
>>>> donald.doherty@brainstage.com
>>>> 412-683-1410
>>>>
>>>> This email message is confidential and/or privileged. It is to  
>>>> be used by
>>>> the intended recipient only. Use of the information contained in  
>>>> this email
>>>> by anyone other than the intended recipient is strictly  
>>>> prohibited. If you
>>>> have received this message in error, please notify the sender  
>>>> immediately
>>>> and promptly destroy any record of this email.
>>>>
>>>
>>>
>>
>>
>>
>> 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 - William.Bug@DrexelMed.edu
>>
>>
>>
>>
>

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 - William.Bug@DrexelMed.edu

Received on Sunday, 13 May 2007 07:31:48 UTC