Greetings GenBank Users,
GenBank Release 159.0 is now available via FTP from the National
Center for Biotechnology Information (NCBI):
Ftp Site Directory Contents
---------------- --------- ---------------------------------------
ftp.ncbi.nih.gov genbank GenBank Release 159.0 flatfiles
ncbi-asn1 ASN.1 data used to create Release 159.0
Close-of-data for GenBank 159.0 occured on 04/12/2007. Uncompressed, the
Release 159.0 flatfiles require roughly 275 GB (sequence files only)
or 288 GB (including the 'short directory', 'index' and the *.txt files).
The ASN.1 data require approximately 232 GB.
Recent statistics for non-WGS sequences:
Release Date Base Pairs Entries
158 Feb 2007 71292211453 67218344
159 Apr 2007 75742041056 71802595
And for WGS sequences:
Release Date Base Pairs Entries
158 Feb 2007 86043478524 19421576
159 Apr 2007 93022691867 23446831
During the 63 days between the close dates for GenBank Releases 158.0 and
159.0, the non-WGS/non-CON portion of GenBank grew by 4,449,829,603 basepairs
and by 4,584,251 sequence records. During that same period, 924,995 records
were updated. An average of about 87,448 non-WGS/non-CON records were added
and/or updated per day.
Between releases 158.0 and 159.0, the WGS component of GenBank grew by
6,979,213,343 basepairs and by 4,025,255 sequence records.
For additional release information, see the README files in either of
the directories mentioned above, and the release notes (gbrel.txt) in
the genbank directory. Sections 1.3 and 1.4 of the release notes
(Changes in Release 159.0 and Upcoming Changes) have been appended
below.
** Important Note **
GenBank 'index' files are now provided without any EST content, and without
most GSS content. See Section 1.3.3 of the release notes for further details.
NCBI is considering ceasing support for the index files, so we encourage
affected users to review that section and provide feedback.
Release 159.0 data, and subsequent updates, are available now via
NCBI's Entrez and Blast services.
As a general guideline, we suggest first transferring the GenBank release
notes (gbrel.txt) whenever a release is being obtained. Check to make sure
that the date and release number in the header of the release notes are
current (eg: April 15 2007, 159.0). If they are not, interrupt the
remaining transfers and then request assistance from the NCBI Service Desk.
A comprehensive check of the headers of all release files after your
transfers are complete is also suggested. Here's how one might go about
this on a unix platform, using csh/tcsh :
set files = `ls gb*.*`
foreach i ($files)
head -10 $i | grep Release
end
Or, if the files are compressed, perhaps:
gzcat $i | head -10 | grep Release
If you encounter problems while ftp'ing or uncompressing Release
159.0, please send email outlining your difficulties to:
info from ncbi.nlm.nih.gov
Mark Cavanaugh, Vladimir Alekseyev, Michael Kimelman
GenBank
NCBI/NLM/NIH/HHS
1.3 Important Changes in Release 159.0
1.3.1 Organizational changes
The total number of sequence data files increased by 76 with this release:
- the BCT division is now comprised of 21 files (+1)
- the CON division is now comprised of 7 files (+3)
- the ENV division is now comprised of 5 files (+1)
- the EST division is now comprised of 604 files (+15)
- the GSS division is now comprised of 252 files (+50)
- the HTG division is now comprised of 93 files (+3)
- the INV division is now comprised of 10 files (+1)
- the PLN division is now comprised of 25 files (+1)
- the PRI division is now comprised of 33 files (+1)
The total number of index files increased by 2 with this release:
- the AUT index is now comprised of 21 files (+1)
- the JOU index is now comprised of 3 files (+1)
1.3.2 Several CON-division files significantly exceed 2GB
Data generated by the AACY02 metagenomic WGS project:
PubMed ID 17355176
http://www.pubmedcentral.nih.gov/articlerender.fcgi?tool=pubmed&pubmedid=17355176
resulted in the addition of over 3 million CON-division records to GenBank.
Current procedures are not optimized for handling such a large influx, and
as a result, several of the CON-division sequence files exceed 2GB when
uncompressed. For example:
gbcon5.seq : 7,261,904,050 bytes
gbcon6.seq : 3,092,551,793 bytes
gbcon7.seq : 5,859,018,078 bytes
During the next several months, we will work to enhance the processing of
the CON-division of GenBank, such that the records are provided via a
larger number of much smaller files.
1.3.3 Changes in the content of index files
As described in the GB 153 release notes, the 'index' files which accompany
GenBank releases (see Section 3.3) are considered to be a legacy data product by
NCBI, generated mostly for historical reasons. FTP statistics of January 2005
seem to support this: the index files were transferred only half as frequently as
the files of sequence records. The inherent inefficiencies of the index file
format also lead us to suspect that they have little serious use by the user
community, particularly for EST and GSS records.
The software that generated the index file products received little
attention over the years, and finally reached its limitations in
February 2006 (Release 152.0). The required multi-server queries which
obtained and sorted many millions of rows of terms from several different
databases simply outgrew the capacity of the hardware used for GenBank
Release generation.
Our short-term solution is to cease generating some index-file content
for all EST sequence records, and for GSS sequence records that originate
via direct submission to NCBI.
The three gbacc*.idx index files continue to reflect the entirety of the
release, including all EST and GSS records, however the file contents are
unsorted.
These 'solutions' are really just stop-gaps, and we will likely pursue
one of two options:
a) Cease support of the 'index' file products altogether.
b) Provide new products that present some of the most useful data from
the legacy 'index' files, and cease support for other types of index data.
If you are a user of the 'index' files associated with GenBank releases, we
encourage you to make your wishes known, either via the GenBank newsgroup,
or via email to NCBI's Service Desk:
info from ncbi.nlm.nih.gov
Our apologies for any inconvenience that these changes may cause.
1.3.4 GSS File Header Problem
GSS sequences at GenBank are maintained in two different systems, depending
on their origin, and the dumps from those systems occur in parallel. Because
the second dump (for example) has no prior knowledge of exactly how many GSS
files will be dumped from the first, it does not know how to number its own
output files.
There is thus a discrepancy between the filenames and file headers for
forty-one of the GSS flatfiles in Release 159.0. Consider gbgss210.seq :
GBGSS1.SEQ Genetic Sequence Data Bank
April 15 2007
NCBI-GenBank Flat File Release 159.0
GSS Sequences (Part 1)
86834 loci, 64349710 bases, from 86834 reported sequences
Here, the filename and part number in the header is "1", though the file
has been renamed as "210" based on the number of files dumped from the other
system. We will work to resolve this discrepancy in future releases, but the
priority is certainly much lower than many other tasks.
1.4 Upcoming Changes
1.4.1 Multiple identifiers for the PROJECT line
The recently-introduced PROJECT linetype (see Section 3.4.7.2) provides a
way to link GenBank sequences that are part of a sequencing project to the
Entrez Genome Project database, where further details about the project can
be obtained.
As of June 2007, multiple identifiers will be valid for the PROJECT line.
Here is a mocked-up example of the expected usage:
LOCUS AANA01000001 2 rc DNA linear BCT 09-FEB-2007
DEFINITION Polaribacter dokdonensis MED152 whole genome shotgun sequencing
project.
ACCESSION AANA01000001
VERSION AANA01000001.1 GI:85822094
PROJECT GenomeProject:13543 GenomeProject:99999
There are several situations in which a record could be considered part
of two different Genome Projects. For example, consider an
environmental-sampling metagenomic WGS project for which the individual
sequence-overlap contigs are not attributed to a specific organism. A
Genome Project could exist that provides further details about the
sequencing effort, the centers involved, etc.
If, in subsequent assembly and annotation phases, scaffold/super-contig/
chromosomal records are created which **are** attributed to a specific
organism, then those CON-division records could have two Genome Project IDs:
one for the WGS sequencing project as a whole; and a second for organism-
specific Genome Projects.
Additional examples illustrating the use of multiple Genome Project IDs
will be provided in future release notes, and via the GenBank listserv.
