[ensembl-dev] get conservation of a region based on multi-species genomic alignment

Julien Roux julien.roux at unil.ch
Thu Jun 11 10:03:44 BST 2015


Hi Matthieu and thanks for your reply
> We have experience with GERP, and we routinely run it on the 39-way 
> and other sets (the ConservationScore object you've found in our API). 
> We don't do it for primates though, so you could use our API to dig 
> conservation between all *mammals*, not only primates.
> The score() method of GenomicAlignBlock returns the LASTZ score of the 
> aligned regions. It's not a measure of conservation
Ok, so if I want to dig up conservation across all mammals for a given 
alignment block, would the following command work?
foreach my $genomic_align_block( @{ $genomic_align_blocks }) {
   my $restricted_gab = 
$genomic_align_block->restrict_between_reference_positions($seq_region_start, 
$seq_region_end);
   my $conservation = 
$conservation_score->genomic_align_block($restricted_gab);
}

> I think the main question is: do you want a score that reflects the 
> conservation of the region amongst the 8 species ? Or something that 
> tells how much the human sequence is shared by the other primates ? 
> (GERP is of the 1st category, for instance).
Actually, rather the second category (how much the human sequence is 
shared by the other primates?)
> Then how do you want to interpret the conservation score ? Similarity 
> between the primates, or presence of an ancestral state ? We could 
> imagine that a human region is not conserved amongst all primates, but 
> still is the same as in the last common ancestor of primates / mammals
That is an interesting option too. Do you infer the the ancestral 
sequences for a genomic alignment block? How to access it?
I could calculate the % similarity between the human sequence and the 
ancestral primate sequence
Cheers
Julin
>
> Matthieu
>
> On 03/06/15 14:52, Julien Roux wrote:
>> Dear list,
>> I am trying to retrieve, for a given human genomic region, the
>> nucleotidic sequence conservation based on the EPO 8-ways genomic
>> alignments.
>> I successfully managed to extract the genomic alignment of a specific
>> human region (see attached script) using help from the tutorial
>> (www.ensembl.org/info/docs/api/compara/compara_tutorial.html), and with
>> the MethodLinkSpeciesSetAdaptor parameters detailed here:
>> http://www.ensembl.org/info/genome/compara/analyses.html
>> Thanks for that!
>>
>> Now, I would like a conservation score for this region. A sensible score
>> could for example be the % of the human nucleotides in this region
>> showing a perfect conservation in the 8 species. Or it could be the
>> average across all nucleotides in this region of the number of species
>> showing the same nucleotide than human. Do you have suggestions, based
>> on your experience on which measure to consider?
>>
>> And before implementing this manually, I would like to be sure that
>> something similar is not already implemented or easy to extract from the
>> API? For example I've seen a score() function for
>> Bio::EnsEMBL::Compara::GenomicAlignBlock objects. What does it
>> correspond to?
>> http://www.ensembl.org/info/docs/Doxygen/compara-api/classBio_1_1EnsEMBL_1_1Compara_1_1GenomicAlignBlock.html#aa1480f9cf9068f9102f67e8b586f1cce 
>>
>> I also found this, but I have no idea if it can be used for genomic
>> alignments:
>> http://www.ensembl.org/info/docs/Doxygen/compara-api/classBio_1_1EnsEMBL_1_1Compara_1_1ConservationScore.html 
>>
>>
>> Thanks for any tip!
>> Best
>> Julien
>>
>> -- 
>> Julien Roux
>> Marie-Curie postdoctoral fellow
>> Department of Ecology and Evolution, University of Lausanne, Switzerland
>> http://www.unil.ch/dee/home/menuinst/people/post-docs--associates/dr-julien-roux.html 
>>
>> Tel: +41 78 700 2931
>>
>>
>>
>> _______________________________________________
>> Dev mailing list    Dev at ensembl.org
>> Posting guidelines and subscribe/unsubscribe info: 
>> http://lists.ensembl.org/mailman/listinfo/dev
>> Ensembl Blog: http://www.ensembl.info/
>>
>

-- 
Julien Roux
Marie-Curie postdoctoral fellow
Department of Ecology and Evolution, University of Lausanne, Switzerland
http://www.unil.ch/dee/home/menuinst/people/post-docs--associates/dr-julien-roux.html
Tel: +41 78 700 2931





More information about the Dev mailing list