ssSNPer: web interface to identify statistically similar SNPs (ssSNPs) in the HapMap
 

Please Note:  this web-interface will typically analyse files up to ~10 Megabytes in size. This limit is primarily due to the web server restrictions we have in place. That is, the server will time out before large files have finished uploading. If users wish to run a ssSNPer analysis on larger files, please try the Linux command line ssSNPer [updated 29/03/07].

This web interface allows users to simply upload a file specifying a test SNP rs# and a file containing HapMap SNP genotype data to determine the number and location of SNPs in the surrounding region which are statistically similar and could therefore produce similar (including more significant) association results.

Statistical similarity is measured by the pairwise linkage disequilibrium (LD) r² coefficient.  The output first includes a graph providing an overview of the statistical similarity between the test SNP and surrounding HapMap SNPs.  Below the graph is a list of the surrounding SNPs which have at least 50% similarity [i.e., r²>=0.5] first sorted from highest to lowest r² and then according to map order [i.e., from pter to qter].  Map distances are given as bp from the test SNP. Clicking on the rs# opens the SNP's dbSNP Cluster Report in a new window.  The final section reports the r² values between the test SNP and all HapMap SNPs across the submitted region in map order; this data is suitable for plotting along with other map landmarks (e.g., genes, conserved regions, etc).

SNPs with r² values of 1.0 are "statistically indistinguishable" (siSNPs) (also previously termed "genetically indistinguishable", giSNPs) and will produce identical association results.  SNPs with r² values >=0.8 can be considered to be of high enough similarity [i.e., SNP 1 has 80% power to predict the genotypes of SNP 2 and vice versa] to produce similar and perhaps more significant association results compared to the test SNP.  Therefore, researchers should consider investigating ssSNPs (especially with r²>=0.8) and their surrounding plausibly causative region(s) in addition to the region(s) initially implicated by the test SNP.

The maximum HapMap genotype data dump file region size is 5 Mb via the HapMap database "Browse Project Data" online web interface.  Although, it is unlikely users will want to examine larger region sizes [i.e., significant  LD will not extend over such distances], the ssSNPer web interface should easily accommodate larger datasets encompassing 10,000s of SNPs.

ssSNPer is also useful for choosing replacement SNPs in the late stages of association study design.  More specifically, the ssSNPs identified can be preferentially chosen with respect to their r² values to replace previously selected SNPs which do not fulfil particular assay design requirements.

Please note: to allow maximum flexibility of user-supplied HapMap data [i.e., either unrelated, trio or a mixture of unrelated and trio data], ssSNPer calculates r² values using ALL individuals and ignoring familial relationships.  This approach may result in negligible differences in r² values (up to ~0.05) calculated from trio data, where only founder genotypes are typically utilised.

Please use the following reference when reporting results obtained via this web interface:

Nyholt DR (2006) ssSNPer: identifying statistically similar SNPs to aid interpretation of genetic association studies. Bioinformatics 22(23):2960-2961.

Some user may also be interested in a novel application of a modified command line ssSNPer which was used to identify Minor Histocompatibility Antigens (mHags), for more details link to http://www.umcutrecht.nl/subsite/dcch/Research/Hemato-Oncology/Identification-of-novel-GvT-associated-minor-H-antigens.htm
 

The ssSNPer interface takes two files as input:

1) a plain [ascii/ansi] text file containing the HapMap SNP rs# you wish to investigate (e.g., named "rs1800630snp.txt").
[NB: this file should contain only ONE line listing the SNP rs# (e.g., "rs1800630").]

2) a HapMap SNP genotype data dump file for the region surrounding the SNP you wish to investigate (e.g., named "rs1800630dumpedregion.txt").
[NB: See below for step-by-step instruction for obtaining this file.]

These files are run through a modified version of Gonçalo Abecasis' LDMAX program - part of the GOLD (Command Line Tools) package [gold-1.1.0.tar.gz].  LDMAX calculates r² values from haplotype frequencies estimated via the expectation-maximization (EM) algorithm of Slatkin and Excoffier (1995), Mol Biol Evol 12:921-7.

The following links show example input files:
rs1800630snp.txt
rs1800630dumpedregion.txt

The following link shows the output obtained using the example input files:
rs1800630results.pdf