How to create 10GB blast database

Question

[arronar]

Hello.

I want to create a database for local blast but fasta file with reads is 9Gb. So I want from this file to create only one database for the blast and not many files. So I put -max_file_sz argument as you can see above, but it creates many files of 2.1GB. How can i create only one database file? Should I merge them later?

Thank you.

Here is the command I use.

makeblastdb -in /mnt/usb/sra/merged.fasta -max_file_sz '10GB' -dbtype nucl -out /mnt/usb/sra/merged_db

Answer 1

[Zhaorong]

I may be wrong, but it seems "-max_file_sz" does not allow a size greater than 2GB.

Checkout makeblastdb source code line #1122-1126:

Uint8 bytes = NStr::StringToUInt8_DataSize(args["max_file_sz"].AsString());

if (bytes >= (1L << 31)) {

    NCBI_THROW(CInvalidDataException, eInvalidInput, "max_file_sz must be < 2 GiB");

}

Links:

1. Uint8
2. NStr
3. StringToUInt8_DataSize
4. AsString
5. bytes
6. NCBI_THROW
7. CInvalidDataException

I think an alternative is to use the blastdb_aliastool that comes with NCBI BLAST+. Please refer to the "Aggregate existing BLAST databases" section of BLAST Command Line Applications User Manual. I copied it below for easier access.

To combine the two nematode nucleotide databases, named "nematode_mrna" and "nematode_genomic", we use the following command line:

$ blastdb_aliastool -dblist "nematode_mrna nematode_genomic" -dbtype nucl \
  -out nematode_all -title "Nematode RefSeq mRNA + Genomic"

[arronar]

Thank you but this doesn't really merge the databases. Instead, it creates a virtual database in a .nal file.

So is there any way to merge .nhr , .nin and .nsq files to have a single only database?