Re: [Treebase-devel] ABI proposal for phyloinformatics

[William P. <wil...@ya...>]

Yeah, since ToLWeb provides an XML dump of the whole tree, from the TreeBASE side a cross-walk with ToLWeb (or mechanisms to benefit from the ToLWeb skeleton for enhancing the searching of TreeBASE) could be done without *any* modifications of ToLWeb (or at least we could provide ToLWeb with clade-level query links that they can add to each clade page if they choose to).  Likewise, we will no-doubt want to provide higher-name-sensitive queries using NCBI and ITS (etc) without requiring permission to change their code (to say the least!). 

But clearly we want something tighter than just consuming ToLWeb's XML -- and we want ToLWeb to benefit as much as TreeBASE. But in that case, I think some sort of tap on David Maddison and Andrew Lenards's shoulder is needed (David is on the phylorf mailing list, but is he following this?).  I gather that ToLWeb code is not yet Open Source... 

bp


On Jun 7, 2011, at 1:48 PM, Hilmar Lapp wrote:

> Jim:
> 
> On Jun 7, 2011, at 1:30 PM, Jim Leebens-Mack wrote:
> 
>> I am a bit concerned about the tight connection between TB and ToLWeb that is outlined in the pitch.  Are folks intending to re-engineer BOTH TB and ToLWeb?  In my mind, ToLWeb would be just one of many platforms from which folks may want to delve into the phylogenetic knowledge that could be accessed in TreeBASE.
> 
> 
> Good points, and indeed what I had in mind technically(*). The way I am envisioning this to be implemented is indeed using technologies (HTTP/REST APIs, canonical resolvable identifiers, RDF) that allow very loose coupling. I left that out as I thought the tech soup shouldn't be in there, but I agree what's missing now is the notion that this will be achieved through loose coupling.
> 
> So as for reengineering, the idea is to engineer (or reengineer where that's necessary) components for *both* systems that allow that loose coupling in a way that achieves the stated goals. Components that need not change to achieve this would not be touched.
> 
> 	-hilmar
> 
> (*)  Socially (as opposed to technically), I think ToLWeb can, and should, play a much more important role in enhancing TreeBASE content in the sense of turning data into knowledge than other platforms that we would enable here. Perhaps a useful analogy to think about is Genbank as the the raw sequence data repository and NCBI Gene (and its predecessor LocusLink) as well as RefSeq as resources that attempt to turn this into curated knowledge.