Tools For Predicting Functional Effects Of Snps?

Question

[viavialin]

Hi Everyone,

I'm new to this field and first time posting, please bear with me.

I am working on locating the functional effects of SNPs which appear to be relevant in our candidate gene study. I am aware of individual websites such as miRanda, TRANSFAC, ESE-finder etc that can help predict miRNA binding sites, TF binding sites and exon splicing elements. However, with a long list of SNPs to search, (eg. 100 or more), it is quite a pain to search each individual site with 1 SNP at a time.

Is there a meta-tool that someone can recommend which allows easy input (rs# only) and combines many functional options?

I have already tried some tools, many of which are no longer working - F-SNP, Pupasuite, SNPit etc, Amongst the others that have given me some results, SNP Function Portal, SNPinfo and pfSNP, I am having trouble 'trusting' them since all of these sites have given me different answers (such as TFs) for the same SNP I tested.

If anyone has some insights, I would be very grateful! Thanks for your patience.

[brentp]

did you try SeattleSeq?

[sci11168]

Hi. I'm the developer of pfSNP. Just to clarify why the results may be different in pfSNP vs SNP Function Portal etc. The first reason is that we used different TF PWM. The 2nd reason is that we may be using different threshold for calling a TF site. pfSNP used a threshold to minimize both false positive and false negative.

[viavialin]

Thanks all for your input! I will try this.

[viavialin]

Yes, I looked at SeattleSeq but the input format is VCF? I only have the SNP rs#.

[NB]

you can also upload a "custom" format in SeattleSeq by specifying the column numbers for chromosome,location,reference and alternate allele. If you have rs ids, you can get this information easily..... or you could also try SNPnexus

[sara]

can you try with snpEff

Answer 1

[sci11168]

Hi. I'm the developer of pfSNP. Just to clarify why the results may be different in pfSNP vs SNP Function Portal etc. The first reason is that we used different TF PWM. The 2nd reason is that we may be using different threshold for calling a TF site. pfSNP used a threshold to minimize both false positive and false negative.

Answer 2

[Jorge Amigo]

if you are able to get the proper variant information from that SNP list (a simple script querying dbSNP would do) you could use any annotator available, such as snpEff, Variant Effect Predictor, SeattleSeq, or the one we find more useful in our local pipelines ANNOVAR. if you don't find this easy enough and you could live with a limited query and a limited amount of information retrieved, if you have a list of SNPs you could always try entering them into Ensembl's BioMart. the idea would be to select Ensembl Variation as database and Short Variation as dataset, enter the rs codes in the Filter by Variation Name box (web query is limited to 500 SNPs), and selecting attributes like PolyPhen or SIFT score.

[viavialin]

Thanks, I'll try this. What type of script do I use to query dbSNP? I'm afraid I don't have much knowledge on writing codes. I'll look into BioMart. Would a tool like SNPinfo or SNP Function Portal be less robust than Ensembl?

[Jorge Amigo]

those 2 tools you mention are quite old. you should always try to find the most complete and updated resource for your needs. Ensembl contains all dbSNP, which is the current reference variation catalog, plus it adds those 2 functional prediction scores to your output. if that's enough, I would choose this option as valid. the best solution would be to get the SNP information (chr, pos, reference allele, variant allele,...; if you don't know how to get that through querying dbSNP - mysql, scripting,... - you can get it again through Ensembl) and feed any of the annotation tools mentioned above to get any of the functional annotations publicly available.