question on making BLAST db

[rdu]

Hi,

I'm trying to make a BLAST database. The fasta file was downloaded from ftp://ftp.theseed.org/genomes/SEED/SEED.fasta.gz. Then I used commond line: makeblastdb -in seed.fasta -out seeddb -dbtype prot. But it shows a error message "Eror: Unable to open input seed.fasta as either FASTA file or BLAST db".

The seed.fasta is around 1.2G big. I opened it, the lines are like:

>fig|100226.1.rna.129 unknown function
......................................................
......................................................

>fig|100226.1.peg.6568 putative membrane protein (GT:3483031)
MKQGKEVENPESRFFAVLLAAA..................................................
........................................................................................
......................................LGRDFSNTSFAAAPAPPAEPGPETPTVRG


What cause the error? Thanks in advance.

[dschika]

First of all: Have you renamed the 'SEED.fasta' to 'seed.fasta'?
Furthermore, it looks like your fasta file isn't complete valid. At the beginning of the file there are some fasta entries without sequence. You could delete them and try again!

[anna_]

Try to use "-hash_index" in your command. I don't exactly the reason, but I think this causes a kind referencing your database file.

anna

[rdu]

Appreciate, dschika and anna.

The mistake I made is I wasn't at the correct directory. After I switched there "makeblastdb" did take actions. But just like what you both pointed out, the "SEED.fasta" isn't completely valid. (Error message showed as Bad Char String some things.)

My question is seems it isn't applicable to manually edit the big size .fasta file (1.2G), because even just a simple click on the file would take some time to response. Any better way to work on fasta file?

I haven't had a chance to try "-hash_index", but I checked the BLAST maual. It works for "Enables the creation of sequence hash values. These hash values can then
be used to quickly determine if a given sequence data exists in this BLAST database." I'm confused, because I haven't had a database by this point.

By the way, something interrupted and made me reply so late, sorry!

[dschika]

I don't know your programmig skills, but I would write a small python or perl script.