Consult the Data Dictionary to determine which table column holds the heterozygosity data or check the dbSNP Main Table.

Be aware that snp_type is no longer in the table because we never used it.

Functional class is located in the SNPContigLocusId table. If you need help deciphering the table, please consult the Data Dictionary mentioned above.

If you are a power user that regularly downloads dbSNP for in-house analysis, you might be interested in the ER (Entity Relationship) diagram, which delineates dbSNP structure, including the relationship between the database tables.

The sequence coordinate data for the XML, ASN.1, .bcp, and the Genotype/genotype_by_gene files were changed from 1-based to 0-based starting with dbSNP build 125. The ASN.1_flat, Chromosome Report, and the web page reports remain 1- based. (6/30/06)

Some submitters did not submit genotype or frequency data to dbSNP in their submissions; therefore, there is no population frequency data for these SNPs. There are approximately 27 million submitted SNPs in dbSNP, and only 3.5 million of those have frequency data associated with them. (1/9/06)

The ASN.1 format for floating point numbers represents the mantissa (the fractional part of a logarithm, to the right of the decimal point) as a whole number that includes the base number as well as its order of magnitude. In your case, the number would be base (410112023353577) and exponent 10^(-15). (6/8/05)

You can find the description in the FTP 00readme file:

KEYWORD       docsum.asn field (Value(s) or definition(s))
SEQ               1. PrimarySequence.attlist.gi
                  2. PrimarySequence.attlist.source
                 (submitter = sequence source is dbSNP submitter)
                 (blastmb = sequence source is NCBI blast database)
                 (xm = sequence source from NCBI Refseq)
                  3. MapLoc.attlist.asnFrom   [../^][MapLoc.attlist.asnTo]
                 (see** below for symbol definitions)
                  4. MapLoc.attlist.locType   (1=insertion; 2=exact;
                  3=deletion; 4=range-insertion; 5=range-exact;
                  6=range-deletion)
                  5. MapLoc.attlist.orient (+ =forward, - =reverse, ? 
                  =unknown)

(08/07/08)

No, we have a large user base for both formats, and will likely support both formats in the future.(6/8/06)

The dbSNP RefSNP page and efetch should provide the same data.

In this particular case, however, a submitted SNP(ss) that contributed to the extra alleles was withdrawn (probably due to strand or sequence error), but there wasn’t time to make corresponding change in the annotation before the build was released. The annotation will be fixed in the next build (b127), which should be released very shortly. (12/18/06)

The new refSNP cluster number is determined by comparing the refSNP cluster creation date with the last build (b126) release date (which was May 4, 2006).

Here is example XML for rs41318644:

<Rs> 
<Rs_rsId>41318644</Rs_rsId>
<Rs_create>
<Rs_create_build>126</Rs_create_build>
<Rs_create_date>2006-11-29 15:30</Rs_create_date>
</Rs_create>

So you are correct in thinking that the rs_create_build_id should be b127. This build_id assignment rule will be fixed for the next build which we plan to release sometime in July. In the meantime, if you are only interested in new refSNPs, could you use the <Rs_create_date> by comparing it with "May 4, 2006"?

Please also note that in b127, all refSNPs (both old and new) have been mapped to the new genome map (NCBI build 36.2) and so new gene annotations are based on this new map. (5/25/07)

The XML file report for the refSNPs lists only submitted SNPs (ss) that occur within refSNP clusters. At the release of build 125, there were ~5 million newly submitted SNPs that did not get into the clustering pipeline in time for the build release, and were not recognized as clustered. Hence, these submitted SNPs did not make it into the XML report. The ~5 million unclustered submitted SNPs have now been clustered and will be released in the next human build.(4/26/06)

The GenoExchange FTP file has not been created for the new build. If you have a specific set of SNPs you are interested in, you can create an XML file with information from the new build by using the Entrez SNP batch query (i.e., dbSNP batch report) and the Genotype report format option.

We have plans to merge the two XML schemas in the near future. As for the schema merge it will be done some time this year. As we have a lot of ongoing projects currently, I’m unable to be specific about when this will be accomplished.(2/5/07)

It is possible to unzip this file, but because it is so large (the unzipped file is 4.3 Gb or 4393989049 bytes), you must use a UNIX machine, or try PKZIP or WinRAR archivers. WinZip does not handle files over 4 Gb. In the meantime, we will split files over 4 Gb into smaller files before we zip them.

The gene information is encoded in the XML for those genes where we are able to map SNP(s) in your organism of interest. docsum_2005 to determine whether the SNP is coding. You can get the coding counts by grouping the coding-synon | coding-nonsynon | reference | together. You can get the non-coding counts by grouping the mrna-utr | intron |splice-site | locus-region | together.

You can also obtain these counts for all genes in the human genome using Entrez SNP. Select the Preview/Index function and then set the Function Class limits at the top of the form, under the Limits link.

When I used the last method, Entrez SNP produced the following result:

#11 Search human[ORGN]
Limits: intron, locus region, mrna utr, splice site
COUNT=1998902
#10 Search human[ORGN]
Limits: coding nonsynon, reference, coding synonymous
COUNT=49522

dbSNP allows many “bad alleles”, because our parser rules include the following:

1.

Accept all IUPAC codes and a “-”.

2.

If an allele is within parentheses, then it is treated as a description of an allele of the variation when a submitter does not know the actual allele, or when the allele bases are longer than 250.

3.

Do not allow common bases in the beginning or ending of all the alleles of a variation. For example, the “NNN” allele was used in variation “NNN/-”. This means a three-base insertion, but the submitter did not know the actual bases in it.

4.

C/Y means a submitter clearly detected a “C” in some sequences, whereas in other sequences, the base detection was indeterminate—the submitter knows that there might be a “C” or a “T”.

We still have some bad alleles that I am in the process of cleaning up, including the two in your list (“+”, “+ T”). There is also a “+G” in variation “+G/-“. These bad alleles involve four SNPs. I will email the submitters for clarification and not allow this type of allele form in future submissions.

“N” has two meanings, depending on the context in which it is used. The first context in which “N” is used is allele frequency. In this context, “N” means “indeterminate frequency”. For example, if a submitter has a sample size of 120 chromosomes, they may submit A=40/C=78/N=2.

The second context in which “N” is used is in SNP FASTA sequence. Here, a variation is represented by the IUPAC letters of A, C, M, G, R, S, V, T, W, Y, H, K, D, B, and N. If the variation is not represented by one of the first 14 letters, then it is considered an “N”. For example: All indels, microsatellites, and named variations are expressed as “N” in SNP FASTA sequences.

Some submitters in the past have used “+” to represent the insertion part of an indel SNP. You could get the real inserted sequence (relative to the deletion) from the variation from the SNP assay. We realize that allowing “+” may confuse users, so we are in the process of substituting the real “insertion” sequence for the “+” currently available and hope to get this done by build 122.

There are two ways to upload the data for all SNPs that have genotype or frequency information.

The easiest way to do this is to upload the Genotype XML GenoExchange files. These files contain all dbSNP Genotype and allele frequency information arranged by chromosome for each organism. For example, the human build 125 genotype data are located in the genotype subdirectory of the human SNP directory. The schema for the GenoExchange files is also located online.

Another way to upload the data for all SNPs that have genotype or frequency information that requires a little more work, is to upload all the dbSNP database files, which are also available on our FTP site. (5/6/06)

The XML files are generated from the database from which the .bcp files are created. There should be no difference between them, apart from some minor formatting. Please let us know if you detect anything unusual.

Population and frequency information is now located in the genotype files for the organism of interest. You can find this file by going to the dbSNP ftp organism directory, select an organism of interest (in this case I will choose “human_9606”), and then select “genotype” from the list of the organism’s subdirectories. (11/4/05)

Some submitters did not submit genotype or frequency data to dbSNP in their submissions; therefore, there is no population frequency data for these SNPs. There are approximately 27 million submitted SNPs in dbSNP, and only 3.5 million of those have frequency data associated with them. (1/9/06)

If a SNP hits once or twice on the same chromosome, it is assigned to a chromosome file; if it hits more than twice or hits on different chromosomes, it goes to ds_chMulti.xml. This was designed to take account of possible fragment redundancy in unfinished parts of the genomic sequence. There is one notable exception, however. When SNPs hit in the pseudo-autosomal region on Y, they are recorded on both the X and Y chromosome files. Also, we track hits to both the reference genome as well as the alternate assemblies and haplotypes. When a SNP hits on several alternate scaffolds, we record it several times but consider only the number of distinct loci when deciding whether to assign it to chMulti.

The SNP you are looking for is in the ds_ch11.xml file of the human XML directory. This SNP maps to the Celera assembly only, so it will not appear in your Entrez search results since Entrez indexes only SNPs that map to the NCBI reference assembly.(8/30/06)

You can use eUtils or dbSNP Batch Query to retrieve to retrieve specific xml record(s) containing submitted SNP (ss) information. You might want to take a look at this eUtils example for rs242, as well as online eUtils tutorial and examples . (12/01/05)

Please see the online short course for using eutils (9/15/06)

The sequence coordinate data for the XML, ASN.1, .bcp, and the Genotype/genotype_by_gene files were changed from 1-based to 0-based starting with dbSNP build 125. The ASN.1_flat, Chromosome Report, and the web page reports remain 1- based. (6/30/06)

We are glad that you are interested in the new VarView. It is still in a very early stage of release. We will be working on providing XML dumps for VarView data in the future. (05/08/08)

The element FlagDesc in the xml lists the flags and their descriptions. (4/8/05)

The correspondence between the XML tags and the dbSNP schema tables is as follows (the “vw” prefix means that it is a view rather than a table):

XML tag                                 Schema Table
------------------------------------    ---------------------------------------
<NSE-rs_refsnp-id>                      SNP.snp_id
<NSE-rs_create-build>                   vwSNP_build.create_build_id
<NSE-rs_update-build>                   vwSNP_build.last_updated_build_id
<NSE-rs_observed>                       ObsVariation.pattern
<NSE-rs_seq-5_E>                        SubSNPSeq5.line (set of rows,
                                           ordered bySubSNPSeq5.line_num) 
<NSE-rs_seq-3_E>                        SubSNpSeq3.line (set of rows,
                                           orderedbySubSNPSeq5.line_num)
<NSE-rs_het>                            SNP.avg_heterozygosity  
<NSE-rsMaploc_asn-from>                 SNPContigLoc.asn_from
<NSE-rsMaploc_asn-to>                   SNPContigLoc.asn_to
<NSE-rsMaploc_physmap-str>              SNPContigLoc.phys_pos
<NSE-rsMaploc_physmap-int>              SNPContigLoc.phys_pos_from
<NSE-rsContigHit_accession>             SNPContigLoc.contig_acc
<NSE-rsContigHit_version>               SNPContigLoc.contig_ver
<NSE-rsContigHit_chromosome>            SNPContigLoc.contig_chr
<NSE-FxnSet_symbol>                     SNP.ContigLocusId.locus_symbol
<NSE-ExchangeSet_dbSNP-build-number>    Not in the database, set externally

dbSNP propagates the INTERIM tag from Entrez Gene. Although the INTERIM tag is not unique, the locus_id for each gene is distinct and is associated with an mRNA transcipt and protein in the NCBI refseq database (and sometimes more than one, in the case of splice variants).

The INTERIM tag is assigned during the human genome annotation process. Think of INTERIM as a flag indicating that genes have been predicted at this locus, based on mRNA and protein models, but have not yet been curated to a known gene symbol.