I have downloaded complete protein sequences of some bacterial genomes in March 2014 from following NCBI FTP site,
ftp://ftp.ncbi.nlm.nih.gov/genomes/Bacteria/
I had a list of NCBI GIs which were accessible in the on-line NCBI interface (www.ncbi.nlm.nih.gov/protein) till mid of 2014, but now when I search them on NCBI, an error appears stating "Record removed" with no reasons. The page provides me a link to obsolete version, but not to updated version which I actually needed. Few of those GIs are;
378699640
537435588
550903479
NCBI Help Center provided me the reason that NCBI has updated its FTP site in August 2014. They have also modified FASTA format with NCBI-GI removed from the header (see following NCBI News)
http://www.ncbi.nlm.nih.gov/news/08-26-2014-new-genomes-FTP-live/
http://www.ncbi.nlm.nih.gov/news/09-17-2014-simple-FASTA-headers-genomes-FTP/
In this scenario, I needed to repeat my analysis with new genomes. So that other people can access my published results.
My question to NCBI users is that how they tackle such situation ? Has anyone else encountered same issue as mine ?
It's a real problem, especially because it's often difficult to determine why a record was removed. I tend to favor Uniprot for this reason - it gets updated frequently but the protein accession IDs are consistently accessible. The link below corresponds to the first GI in your list as an example.
Of course, Uniprot also deletes entries (see the second link).
http://www.uniprot.org/uniprot/E1WG01
http://www.uniprot.org/help/deleted_accessions
It's a real problem, especially because it's often difficult to determine why a record was removed. I tend to favor Uniprot for this reason - it gets updated frequently but the protein accession IDs are consistently accessible. The link below corresponds to the first GI in your list as an example.
Of course, Uniprot also deletes entries (see the second link).
http://www.uniprot.org/uniprot/E1WG01
http://www.uniprot.org/help/deleted_accessions
Honestly, what you did, by repeating the analysis is really the best way to do this. Keep in mind that all of these genomes are provisional and are constantly being corrected, amended and so forth. When an erroneous entry based on ORF analysis is found, it will be replaced with a corrected version. Really, that is as it should be, as these are works in progress, not static references (and hence why UCSC created the liftover tool for genomic coordinate mapping from one reference genome release to the next). It is a constant issue and inherent in the fact that no reference genome is a wholly verified, static entry. Genomes change, ontologies change, etc - if they did not, we would all be out of a job and sitting at home now :-)
One other suggestion I have is just stop using the GI numeric accession and instead use the RefSeq accession, as that will be more tractable in the long run. Note though that refseq entries do change as they progress from provisional to higher status in the curation process (see http://www.ncbi.nlm.nih.gov/books/NBK21091/table/ch18.T.refseq_accession_numbers_and_mole/?report=objectonly ).
Just an addendum to the aforementioned wonderful suggestions:
Now and then keep a track of the GIs using Batch Entrez tool of NCBI. People seldom use this and I wonder why. It is the best batch analysis tool which will allow you to identify the obsolete records easily. Now please don't stop there, you could annotate the entries and put NCBI a word of your annotation.
You tend to observe many of these with unique Gene Indices (GIs) with hypothetical proteins or unknown. They would certainly be useful for your database to make a good history of negative dataset.
Best
Prash
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools