[go] ISA/ISO

[Kara Dolinski]

Hi,

These are all good points, and I've had some useful, interesting  
offline exchanges with people about some of these issues.

Certainly the current plan in place uses the evolutionary definition  
of the term as described by Michelle, and will of course rely on  
sequence similarity (OrthoMCL starts with BLAST) to find the  
orthologous relationships.    I think/hope everyone would agree that  
this is a good and necessary first step.  But, Val makes a good  
point.  For the purposes of GO annotation, perhaps the bigger picture  
is that we do not want to identify just putative orthologs but also  
genes that are likely share the same function (many are orthologs,  
but many are not, as Val points out in a subsequent email).  If  
that's the case, we'd want to explore alternative methods that look  
at things like conservation of interactions, gene expression, etc  
rather than sequence similarity.  This is, of course, a much bigger  
job, and, not to say it shouldn't be done, but we might want to stay  
off the bleeding edge a bit longer and start with the simpler  
sequence based approach as a start, keeping in mind the caveats that  
Val raises.


On Feb 21, 2008, at 11:50 AM, Pankaj Jaiswal wrote:

> This seems like a case for 'functional ortholog' based on the  
> conserved assembly and interaction among subunits of a protein  
> complex from two different species. We see these cases quite often.
>
> Here is an interesting paper on this theme
> Systematic identification of functional orthologs based on protein  
> network comparison. Genome Res. 2006 Mar;16(3):428-35.
> PMID: 16510899
> http://www.ncbi.nlm.nih.gov/pubmed/16510899?dopt=Abstract
>
> Pankaj
>
>
> Gwinn Giglio, Michelle wrote:
>> Hi Val,
>> The definition of ortholog is that the two sequences arose from a  
>> common
>> ancestor and were separated by a speciation event.  The situation you
>> describe sounds to me like there is good evidence that the two  
>> proteins
>> carry out the same functions, however they could either be  
>> orthologs that
>> have diverged so much that no recognizable sequence similarity  
>> remains or
>> they have arisen due to convergent evolution and are NOT true  
>> orthologs.
>> Michelle
>> On 2/21/08 8:54 AM, "Valerie Wood" <val at sanger.ac.uk> wrote:
>>> In support of the final sibling arrangement of ISA/ISO, I came  
>>> across an
>>> example today where gene products are known to be orthologous but  
>>> have
>>> no sequence conservation which I thought I would share.
>>>
>>> Identification of the Proteins, Including MAGEG1, That Make Up the
>>>  Human SMC5-6 Protein Complex
>>> Elaine M. Taylor, Alice C. Copsey, Jessica J. R. Hudson, Susanne  
>>> Vidot,
>>> and Alan R. Lehmann
>>> quote
>>> "Nse6 contains ARM/HEAT repeats, but there is no sequence  
>>> conservation
>>> between these presumed orthologs from S. cerevisiae/ and /S.  
>>> pombe/. In
>>> /S. pombe/ they also appear to bridge the head domains of Smc5  
>>> and Smc6"
>>>
>>> where it is not even possible to detect the similarity between the 2
>>> nse6 subunits in yeast and pombe but they are presumed orthologs  
>>> from
>>> other biological properties and position in the complex, binding
>>> partners etc.
>>> These are NOT reciprocal best hits, and although they can be  
>>> aligned,
>>> (you can align anything) it does not mean, that if I inferred any
>>> experimental data from the S. cerevisiae ortholog it would come  
>>> from a
>>> sequence alignment.
>>>
>>> I should also add that this is very common which also has large
>>> implications for the orthology prediction exercise.
>>>
>>>
>>> Val
>>>
>>>
>
> -- 
> Pankaj Jaiswal
> G-15, Bradfield Hall
> Dept. of Plant Breeding and Genetics
> Cornell University
> Ithaca, NY-14853, USA
>
> Ph. +1-607-255-3103 / 4199
> fax: +1-607-255-6683
>