SNP function prediction in non-human species

[spark]

Hello,

I'm trying predict SNP function for a few thousand non-synonymous coding sequence SNPs in cattle. I've looked through the documentation for SIFT and PolyPhen, but can't easily see if they can be configured to work with bovine genome data. Does anyone have any experience of using these or other tools for SNP function prediction in non-human species?

Thanks,

Stephen

[jeny]

I am not sure if it could do what you want, but have you tried variant_effect_predistor of Ensembl http://www.ensembl.org/tools.html?

[spark]

Thanks for the suggestion. I've since used the Ensembl variant effect predictor, and it does something very similar to Annovar (which I've already used). The only difference is that the Ensembl program also outputs the amino acid residues for exonic SNPs.

What neither program does (and what I would like to do) is to use multiple sequence alignment data from many taxa to identify which SNPs affect highly conserved amino acid positions. Only tools like Sift and PolyPhen appear to do this, and I still haven't found a way to apply them to large batches of non-human data.

.

[clarissaboschi]

Spark did you solve the problem?
I want to do the same, I installed the SIFT in Unix, but I did not find any information how should be the input file and command line. I know it is possible to perform with SIFT in Linux many sequences, but I don’t know how to do it.

I noticed that there are few databases when you install the SIFT for human, cattle, chicken, etc. So I think we need a list of all protein sequences with the substitutions, but I am not sure and I dont know what are the command lines...

Thanks

[shoegame2001]

Does anyone know if there is a SNP effect predictor for variants discovered in cDNA sequences for non-model organisms? I did a de novo transcriptome assembly and discovered variants in those data. No reference genome exists for my species. I can sort out synonymous vs. non-synonymous SNPs using BLAST to related species' proteins, but I was hoping for some more sophisticated prediction of the variant effect.

Given the discussion above, it sounds like it might not exist.

[Dameon]

Try using snpEff. If you have a reference_genome.fasta and either a .gff or .gtf file of the CDS regions, you can build your own custom database for SNP effect prediction. In the case of bovine, if the reference exists in Ensembl, the database can be easily downloaded in the latest version of snpEff.

http://snpeff.sourceforge.net/download.html

[shoegame2001]

Okay, thanks. But no options if you don't have a reference genome?

[Robby]

Is there any tool, which predicts for each variant the probability, that this variant is damaging (like SIFT or PolyPhen for human). As far as I understand snpEff it just shows you the region of the variant and the effect like i.e. aminoacid-change, but not the damaging prediction score. I found also Mouse SNP Miner, but it seems, that this tools works just for known variants.