add extra data loader; update intron exon loader and add util to retreive human refse...

diff --git a/utils/intron_exon_loader.pl b/utils/intron_exon_loader.pl
index 99ace9de358d5aac71083e4d00db7f414ef5f0c9..fef9dca7c919daacb2d06815f28a6382ee315005 100755 (executable)
--- a/utils/intron_exon_loader.pl
+++ b/utils/intron_exon_loader.pl
@@ -46,11 +46,11 @@ Display this manual.
 
  zcat CHR_*/hs_ref_GRCh37.p2_*.gbs.gz | \
  intron_exon_loader| \
- psql snp -c "COPY mrna FROM STDIN WITH DELIMITER '   ' NULL AS 'NULL'";
+ psql snp -c "COPY mrna FROM STDIN WITH DELIMITER '    ' NULL AS 'NULL'";
 
  zcat CHR_*/hs_ref_GRCh37.p2_*.gbs.gz | \
  intron_exon_loader --output cds| \
- psql snp -c "COPY cds FROM STDIN WITH DELIMITER '   ' NULL AS 'NULL'";
+ psql snp -c "COPY cds FROM STDIN WITH DELIMITER '     ' NULL AS 'NULL'";
 
 
 =cut

diff --git a/utils/load_extra_data.sh b/utils/load_extra_data.sh
new file mode 100644 (file)
index 0000000..182ffc0
--- /dev/null
+++ b/utils/load_extra_data.sh
@@ -0,0 +1,25 @@
+#!/bin/sh
+
+DATA_DIR=/srv/ncbi/refseq/
+SCHEMA_DIR=/srv/ncbi/db_snp_utils/schema
+UTIL_DIR=${SCHEMA_DIR}/../utils/
+
+
+psql snp < ${SCHEMA_DIR}/extra_schema/mrna_cds_table.sql
+psql snp < ${SCHEMA_DIR}/extra_schema/intron_exon_schema.sql
+
+zcat ${DATA_DIR}/human.rna.gbff.gz | \
+    ${UTIL_DIR}/human_mrna_cds_insert.pl | \
+    psql snp -c "COPY mrna_cds_table FROM STDIN WITH DELIMITER '       ' NULL AS 'NULL'";
+
+zcat ${DATA_DIR}/seq_contig.md.gz | \
+    ${UTIL_DIR}/import_seq_component_md.pl | \
+    psql snp -c "COPY contigs FROM STDIN WITH DELIMITER '      ' NULL AS 'NULL'";
+
+zcat ${DATA_DIR}/CHR_*/hs_ref_GRCh37.p*_*.gbs.gz | \
+    ${UTIL_DIR}/intron_exon_loader.pl --type mrna| \
+    psql snp -c "COPY mrna FROM STDIN WITH DELIMITER ' ' NULL AS 'NULL'";
+
+zcat ${DATA_DIR}/CHR_*/hs_ref_GRCh37.p*_*.gbs.gz | \
+    ${UTIL_DIR}/intron_exon_loader.pl --type cds| \
+    psql snp -c "COPY cds FROM STDIN WITH DELIMITER '  ' NULL AS 'NULL'";

diff --git a/utils/retreive_human_refseq.sh b/utils/retreive_human_refseq.sh
index 60440b81f358806544ea9b784ff91cfb4b5dea35..a7d6d78e5c6b5a1ac42f6383da7e779a8b7f7e9a 100755 (executable)
--- a/utils/retreive_human_refseq.sh
+++ b/utils/retreive_human_refseq.sh
@@ -1,4 +1,5 @@
 #!/bin/sh
 
 rsync -rvP -m --include '**GRCh37.p*_*.gbs.gz' --include '**GRCh37.p*_*.fa.gz' --include '**/' --exclude '*' rsync://ftp.ncbi.nlm.nih.gov/genomes/H_sapiens/CHR* .
-rsync -rvP rsync://ftp.ncbi.nlm.nih.gov/refseq/H_sapiens/mRNA_Prot/human.rna.gbff.gz .
\ No newline at end of file
+rsync -rvP rsync://ftp.ncbi.nlm.nih.gov/refseq/H_sapiens/mRNA_Prot/human.rna.gbff.gz .
+rsync -LvP rsync://ftp.ncbi.nlm.nih.gov/genomes/H_sapiens/mapview/seq_contig.md.gz .