Local blast search against nt db gains less results than online search

Question

[john]

For my current project I am using blastn and a local version of the nt data base to search homologous sequences. After some time I came to realize that my local searches lack allot of hits that I can get with the same input sequence using the online service.

My local command looks like so.

blastn -db nt -query input.fasta -max_target_seqs 10000 -outfmt "6 evalue bitscore pident sseqid slen sstart send sframe qstart qend qframe qseq sseq staxid" -out blastn.results

The only changes I did to the default parameters in the online and the local version is setting the "max_target_seqs" to 1000.

I will find a couple of sequences in the online version which have higher e-values than my worst hits in the local search and for which I can confirm with blastdbcmd that they are present in my local nt db. As an example I would have loved to found the sequence OU342721.1 the Longspined bullhead. But this hit is not present in my local results yet in the online version. I tried to confirm that my local nt database contains the sequence like so:

$ blastdbcmd -db nt -outfmt %T -entry OU342721.1
61643

The lowest e-value I retrieve in my local search is 4.60e-08 in the online version 2e-30. So there is a many more sequences missing in my local search.

Infos about my local nt database:

$ blastdbcmd -info -db nt

Database: Nucleotide collection (nt)
76,140,253 sequences; 583,847,353,310 total bases

Date: Nov 12, 2021 10:18 AM Longest sequence: 99,791,824 bases

BLASTDB Version: 5

Volumes:
...

[Dunois]

See:

Why the local blast and online blast produce different results?

Cannot reproduce BLAST results returned by the WEB API on the shell-based BLAST+ 2.7.1

[Mensur Dlakic]

In addition to the very useful links that have been offered, your database is about 3 months behind NCBI.

Date: Nov 12, 2021 10:18 AM Longest sequence: 99,791,824 bases

Answer 1

[john]

Okay so i figured it out. The problem is that the default for blastn is to perform a megablast search (set the word size to 28 among others).

So adding the argument -task blastn made the results quite similar.

One usefull trick I found here.

Adding the argument -export_search_strategy strategy_standalone.asn to the local call and using the option save search on the website, provides you with two text files with all the settings. So both searches can be easily compared.