Mouse SNP Miner: an annotated database of mouse functional single nucleotide polymorphisms

Abstract

Background: The mapping of quantitative trait loci in rat and mouse has been extremely successful in identifying chromosomal regions associated with human disease-related phenotypes. However, identifying the specific phenotype-causing DNA sequence variations within a quantitative trait locus has been much more difficult. The recent availability of genomic sequence from several mouse inbred strains (including C57BL/6J, 129X1/SvJ, 129S1/SvImJ, A/J, and DBA/2J) has made it possible to catalog DNA sequence differences within a quantitative trait locus derived from crosses between these strains. However, even for well-defined quantitative trait loci (<10 Mb) the identification of candidate functional DNA sequence changes remains challenging due to the high density of sequence variation between strains.

Description: To help identify functional DNA sequence variations within quantitative trait loci we have used the Ensembl annotated genome sequence to compile a database of mouse single nucleotide polymorphisms (SNPs) that are predicted to cause missense, nonsense, frameshift, or splice site mutations (available at http://bioinfo.embl.it/SnpApplet/). For missense mutations we have used the PolyPhen and PANTHER algorithms to predict whether amino acid changes are likely to disrupt protein function.

Conclusion: We have developed a database of mouse SNPs predicted to cause missense, nonsense, frameshift, and splice-site mutations. Our analysis revealed that 20% and 14% of missense SNPs are likely to be deleterious according to PolyPhen and PANTHER, respectively, and 6% are considered deleterious by both algorithms. The database also provides gene expression and functional annotations from the Symatlas, Gene Ontology, and OMIM databases to further assess candidate phenotype-causing mutations. To demonstrate its utility, we show that Mouse SNP Miner successfully finds a previously identified candidate SNP in the taste receptor, Tas1r3, that underlies sucrose preference in the C57BL/6J strain. We also use Mouse SNP Miner to derive a list of candidate phenotype-causing mutations within a previously uncharacterized QTL for response to morphine in the 129/Sv strain.

Figure 1

Mouse SNP Miner database structure. The core database consists of a MySQL relational database containing information associated with predicted functional mouse SNPs from a selected set of mouse inbred strains. A web-based Java Applet module allows querying, visualization, and downloading of information from the database. Basic information about SNP sequence, location, functional consequence, and associated transcript are derived from private and public sequencing efforts via the dbSNP mouse polymorphism collection mapped onto the annotated Ensembl genome. Additional SNP information was extracted from the OMIM, Symatlas, and GO databases. GO clustering by GeneMerge is queried directly by the Applet viewer prior to downloading. PolyPhen assessment of missense mutation consequence was based on Nrdb orthologous protein alignments, PDB structure information, and protein functional annotation from Uniprot. PANTHER assessment of missense mutation consequence was based on a set of HMM protein alignments. Bold font and arrows pointing out of the database indicate the existence of direct web links from our database to associated database entries.

Figure 2

Web-based access to Mouse SNP Miner database. (not shown) 'SNP Query' mode allows selection of strain or strain group for comparison. Searching can be constrained by chromosomal interval, gene name, SNP accession number, and presence of human ortholog in OMIM database. QTL from the MGI database can be searched by name or keyword and associated chromosome intervals imported for convenient screening. (shown) 'SNP View' mode presents results from the search in a graphical format for convenient run-time scanning. SNPs in the interval are listed by functional consequence and PolyPhen/PANTHER prediction in the upper left and can be rapidly added or removed by clicking on the associated box. Boxes indicating transcripts and lines indicating SNPs are color and symbol coded by functional consequence in the graphical display. The placement of marks above and below the bar indicates transcripts in the forward and reverse strand, respectively. Clicking on a SNP causes detailed SNP information to be displayed in the 'Details' window above. The 'Associations' window displays GO, OMIM, and PolyPhen/PANTHER information and links for the selected SNP. Movement across the chromosome and between SNPs is facilitated by buttons at the bottom of the graphical display. In the example shown, a search has been performed for putative functional SNPs differing between C57BL/6J and all 129 strains (129% allows global searching of 129 strains) for the interval 153,482,802–154,678,264 bp on chromosome 4. A deleterious mutation (Ile706Thr) in the fourth transmembrane domain of Tas1r3 is highlighted.