1 [[!meta title="Postgresql dbsnp Mirror"]]
6 Assuming you are interested in humans (like I am) the following will
7 download the databases and schemas for the current release of dbsnp.
8 For humans, this is currently ≈60G, and takes a while to retrieve
9 (about 24 hours or so).
11 for a in {organism,shared}_{data,schema}; do
12 lftp -c "open ftp.ncbi.nlm.nih.gov; cd /snp/organisms/human_9606/database/; mirror $a";
19 If you're loading the human databases, I've already done the work for
21 [git repository](http://git.donarmstrong.com/dbsnp.git) (`git init
22 db_snp_utils; git pull http://git.donarmstrong.com/dbsnp.git`).
24 Otherwise, you'll want to use
25 [mssql_psql_conversion.pl](http://git.donarmstrong.com/?p=dbsnp.git;a=blob;f=utils/mssql_psql_conversion.pl;hb=HEAD)
26 in my dbsnp git repository to convert your organism's schema into
27 appropriate bits. Something like the following will get you close (you
28 may still need to tweak the sql manually):
30 for a in {organism,data}_schema/*.sql; do
31 ./mssql_psql_conversion.pl ${a} > ${a%%.sql}_postgrseql.sql;
37 Once the schema are correct, you want to load the schema and the data,
38 then apply the constraints and indexes. This will take some time even
39 on a fairly fast machine.
40 [I would expect at least 2-3 days, unless you have exceptionally fast disks.]
42 I have included a script in the utils directory of the git repository
44 [load_snp_data.sh](http://git.donarmstrong.com/?p=dbsnp.git;a=blob;f=utils/load_snp_data.sh;hb=HEAD),
45 which applies the schema, loads the data, and then applies the indexes
46 and constraints. It looks like the following:
48 psql -c 'DROP DATABASE snp';
49 psql -c 'CREATE DATABASE snp';
51 DATA_DIR=/srv/ncbi/db_snp/
52 SCHEMA_DIR=/srv/ncbi/db_snp_utils/schema
53 UTIL_DIR=${SCHEMA_DIR}/../utils/
55 (cd ${SCHEMA_DIR}/shared_schema;
56 cat dbSNP_main_table_postgresql.sql |psql snp;
58 (cd ${SCHEMA_DIR}/human_9606_schema;
59 cat *_table_postgresql.sql|psql snp;
60 ${UTIL_DIR}/human_gty1_indexes_creation.pl create trigger |psql snp;
62 (cd ${DATA_DIR}/shared_data;
63 for a in $(find -type f -iname '*.bcp.gz' -printf '%f\n'|sort); do
65 zcat $a | perl -pe 's/\r/\\r/g' |psql snp -c "COPY ${a%%.bcp.gz} FROM STDIN WITH NULL ''";
68 (cd ${DATA_DIR}/organism_data;
69 for a in $(find -type f -iname '*.bcp.gz' -printf '%f\n'|sort); do
71 zcat $a | perl -pe 's/\r/\\r/g' |psql snp -c "COPY ${a%%.bcp.gz} FROM STDIN WITH NULL ''";
74 (cd ${SCHEMA_DIR}/shared_schema;
75 cat dbSNP_main_index_postgresql.sql dbSNP_main_constraint_postgresql.sql|psql snp;
77 (cd ${SCHEMA_DIR}/human_9606_schema;
78 cat *_{index,constraint}_postgresql.sql|psql snp;
79 ${UTIL_DIR}/human_gty1_indexes_creation.pl index |psql snp;
85 Once the process above has finished, you can actually query the
86 database. For example, to select information about rs17849502 with its
87 chromosome and position, you do something like the following:
89 SELECT scpr.snp_id AS snp_id,
94 uv.var_str AS var_str,
95 ruv.var_str AS rev_var_str,
97 ml.locus_symbol AS symbol,
98 gitn.gene_name AS description
100 JOIN b135_snpchrposonref_37_3 scpr ON s.snp_id=scpr.snp_id
101 JOIN b135_snpcontigloc_37_3 scl ON scpr.snp_id=scl.snp_id
102 JOIN b135_contiginfo_37_3 ci ON scl.ctg_id = ci.ctg_id
103 LEFT OUTER JOIN b132_snpcontiglocusid_37_1 ml ON s.snp_id=ml.snp_id
104 LEFT OUTER JOIN geneidtoname gitn ON ml.locus_id=gitn.gene_id
105 JOIN univariation uv ON s.univar_id=uv.univar_id
106 JOIN univariation ruv ON uv.rev_univar_id=ruv.univar_id
107 WHERE ci.group_term LIKE 'GRCh%' AND s.snp_id='17849502';
109 [I personally use this very query in a program called snp_info, which I'll probably share later.]