BLAST+ creating custom blast database and using blast+ filtering features

deniz · 10-05-2010, 11:56 PM

Hi,

I would like to create a personal blast database of arbitrary sequences and be able to use all the features of BLAST+ to create subsets of databases based on identifiers or filter based on taxonomy.

It looks like the formatting of the definition line in the input FASTA files is crucial to assign proper sequence identifiers.

Using the General database identifier gnl|database|identifier or local identifier format lcl|identifier I wasn't able to use the blastdb_aliastool to create db subsets as it expects a GI list as input. I also didn't have any luck assigning taxonomy identifiers with the -taxid_map option of makeblastdb.

What is the recommended way to format FASTA definition lines in order to be able to use all the filtering features of the BLAST+ tools.

I was thinking of creating pseudo GenBank definitions for all my sequences: gi|<gi-number>|gb|<AccessionVersion>|<Accession>, where <gi-number> is a generated numeric value, and <Accession/Version> is my identifier. This works for the GI based filtering, however it seems like an ugly hack and I would prefer something more straight forward.

How is the taxid_map file formatted? I've tried <gi-number>, <gb|<AccessionVersion>>, or <gb|Accession> as the sequence identifier, however they don't seem to be assigned properly and blastdbcmd with -outfmt %T gives me zero for all entries.

Thanks for any help,

Deniz

PS: I've posted this already in http://I.SEQanswers.com but as it's still beta traffic is a bit low and I'm probably more likely to get an answer here in the forum.

nupurgupta · 06-14-2012, 07:43 AM

Hi,
I was wondering if you found a solution for this? I have been having all kinds of problems creating subset blast databases.
Thanks

couttsbr · 10-26-2012, 12:04 PM

GI lists are essentially simple to create, if you have experience with bash, perl, or any programming language that you can use to format text, you may be able to automatically pull the GIs from these files and put them into a GI list. A GI list is simply a text file with one number per line, and each number is a GI. There may be some utilities that do this automatically (i.e. FASTA->GI list), though I don't know of them.

Alternatively, a nice way I've found for creating GI lists based on queries is to query NCBI for the data you want. You can do a search in NCBI, click send to on the top-right corner, and export it as a GI list. This is a nice, easy way of getting a GI list for blast subsets, though it's difficult to automate.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
How to create a BLAST database	aliealexandre	Bioinformatics	22	02-02-2015 04:34 AM
Custom local blast results	detq182	Bioinformatics	2	11-04-2014 04:53 AM
BLAST+ vs BLASTALL (legacy BLAST)	Symphysodon	Bioinformatics	4	10-25-2011 03:52 PM
BLAST database error - when changing to new BLAST+ local program	biobio	Bioinformatics	4	06-15-2011 06:20 AM
Database of BLAST	CarlElit	Bioinformatics	1	01-04-2010 07:23 AM

10-05-2010, 11:56 PM	#1
deniz Junior Member Location: Australia Join Date: Sep 2008 Posts: 1	BLAST+ creating custom blast database and using blast+ filtering features Hi, I would like to create a personal blast database of arbitrary sequences and be able to use all the features of BLAST+ to create subsets of databases based on identifiers or filter based on taxonomy. It looks like the formatting of the definition line in the input FASTA files is crucial to assign proper sequence identifiers. Using the General database identifier gnl\|database\|identifier or local identifier format lcl\|identifier I wasn't able to use the blastdb_aliastool to create db subsets as it expects a GI list as input. I also didn't have any luck assigning taxonomy identifiers with the -taxid_map option of makeblastdb. What is the recommended way to format FASTA definition lines in order to be able to use all the filtering features of the BLAST+ tools. I was thinking of creating pseudo GenBank definitions for all my sequences: gi\|<gi-number>\|gb\|<AccessionVersion>\|<Accession>, where <gi-number> is a generated numeric value, and <Accession/Version> is my identifier. This works for the GI based filtering, however it seems like an ugly hack and I would prefer something more straight forward. How is the taxid_map file formatted? I've tried <gi-number>, <gb\|<AccessionVersion>>, or <gb\|Accession> as the sequence identifier, however they don't seem to be assigned properly and blastdbcmd with -outfmt %T gives me zero for all entries. Thanks for any help, Deniz PS: I've posted this already in http://I.SEQanswers.com but as it's still beta traffic is a bit low and I'm probably more likely to get an answer here in the forum.

06-14-2012, 07:43 AM	#2
nupurgupta Member Location: New Jersey Join Date: Aug 2010 Posts: 29	Hi, I was wondering if you found a solution for this? I have been having all kinds of problems creating subset blast databases. Thanks

10-26-2012, 12:04 PM	#3
couttsbr Junior Member Location: Ottawa Join Date: Oct 2012 Posts: 2	GI lists are essentially simple to create, if you have experience with bash, perl, or any programming language that you can use to format text, you may be able to automatically pull the GIs from these files and put them into a GI list. A GI list is simply a text file with one number per line, and each number is a GI. There may be some utilities that do this automatically (i.e. FASTA->GI list), though I don't know of them. Alternatively, a nice way I've found for creating GI lists based on queries is to query NCBI for the data you want. You can do a search in NCBI, click send to on the top-right corner, and export it as a GI list. This is a nice, easy way of getting a GI list for blast subsets, though it's difficult to automate.