After many many hours testing and searching on Internet, I can't make a database alias using the blastdb_aliastool.
It starts with a very undocumented NCBI tool at https://www.ncbi.nlm.nih.gov/books/NBK569848/ (updated in 2021) giving this simple command line
blastdb_aliastool -db nematode_mrna -gilist c_elegans_mrna.gi -dbtype nucl -out c_elegans_mrna -title "C. elegans refseq mRNA entries"
I'm trying it both on Windows and Linux and the problem is the same. Shortly :
Point 4. is done with a command like:
blastdb_aliastool -db nr/nr -out nr/actinop -gilist D:/DEV/var/temp/actinomadura.gi -dbtype prot -title "nr Actino prot"
The result is:
Converted 98631 GIs from D:/DEV/var/temp/actinomadura.gi to binary format in nr/actinop.p.gil
BLAST Database error: BLASTDB alias file creation failed. Some referenced files may be missing
The folder where the Blast executables are placed contains the .ncbirc (Linux) and ncbi.ini (Windows) files where the parameter BlastBD is filled. Each NCBI database is in its own folder, it's why -db is nr/nr and not simply nr. But I found that the .gil file is not created if I don't give the complete path names:
blastdb_aliastool -db f:/BlastDB/nr/nr -out f:/BlastDB/nr/actinop -gilist D:/DEV/var/temp/actinomadura.gi -dbtype prot -title "nr Actino prot"
I read many old posts saying that GI list will disappear, it seems not (we can download it from NCBI and the argument of last tool version has gilist argument), I tried with a list of accession ID instead of GI list but I didn't find which arguments of blastdb_aliastool can work for this way.
The steps look like it's a simple process but I need your help to make it, thanks!
Unfortunately current nr pre-formatted database files available from NCBI do not appear to have gi numbers (I checked a bunch of sequences and you only get N/A). So this mode of blastdb_aliastool is not going to work.
Your alternate option is to pull the sequences you need using blastdbcmd and then make a new database using makeblastdb from that file.
Thank you. How do you checked the GI with the N/A response?
Is there a way to alias using accession numbers? Or taxID? I need to make a database from a list of organisms (some virus, bacterias and fungis). The more I can read from the installed nr, the less I have to go manually on NCBI website to download stuff.
You can use blastdbcmd to retrieve data from pre-formatted databases. An example that is supposed to retrieve gi numbers is the command
$ blastdbcmd -db nr -outfmt %g -entry all
this only returns N/A for entries. You can get accession numbers by replacing %g with %a, which works fine.
If you know the taxID's of organisms you are interested in you can use that information to retrieve sequences with -taxids or -taxidlist options with blastsbcmd. For example following command retrieves sequences for Actinomadura (taxID 1988) from preformatted nr database in fasta format.
$ blastdbcmd -db nr -outfmt %f -taxids 1988
Thank you. It is working now with:
blastdbcmd -db f:/BlastDB/nr/nr -taxids 1988 -out f:/BlastDB/extract/acti3.fa
But it displays this line even if the extraction succeed:
The -taxids command line option requires additional data files. Please see the section 'Taxonomic filtering for BLAST databases' in https://www.ncbi.nlm.nih.gov/books/NBK569839/ for details.
I have taxdb.btd, taxdb.bti and taxonomy4blast.sqlite3 in the nr folder
I found the problem!
I manage the DB folder using .ncbirc on Linux and ncbi.ini on WIndows, so BLASTDB is well defined. In my Windows case BLASTDB=f:/BlastDB.
To be able to use this file, I need to go into the folder (cd command) then launch from there.
To easily work with the data, each database is put on its own folder:
f:/BlastDB/nr/nr.pal (... and other nr files)
f:/BlastDB/swissprot/swissprot.pal (... and other swissprot files)
f:/BlastDB/taxdb/taxdb.btd, taxdb.bti, taxonomy4blast.sqlite3
As I struggled with the missing taxonomy files, I put the 3 taxdb files in /nr folder or in the blastdbcmd folder with the same error.
The solution is to put taxonomy4blast.sqlite3 in the directory specified in the configuration file, so directly in BLASTDB
So having f:/BlastDB/taxonomy4blast.sqlite3 (other files are not necessary) makes the error message disappear.
And I don't need to specify it for -db:
blastdbcmd -db nr/nr -taxids 1988 -out f:/BlastDB/extract/acti3.fa
Log in to answer this question.
