Using BLAST Alignment with low sensitivity

Question

[midox]

Hello,

I would like to use BLAST with Low sensitivity(using DNA sequences). Wha't the best paramters should i choose? this is my command:

blastn -query query.fa -subject subject.fasta -out resultatblastn_cerevisae_real  -reward 5 -penalty -4 -gapopen 10 -gapextend 6 -evalue 1

should i change my parameters?

thanks for your help

[h.mon]

What is your intent? By "low sensitivity" you mean finding only very similar sequences?

[midox]

no, it's the opposite. I want to find sequences with low similarity.

[h.mon]

You want a high sensitivity, low specificity blast, then.

[GenoMax]

Hard to figure out exactly what you want but if it is not what @h.mon says then ..

relax the limits, set penalties to low values and collect all "hits" with -evalue set to 10. Then post-parse the results to select "hits" that fit your criteria for "low similarity".

Answer 1

[Andrzej Zielezinski]

If you want to find DNA sequences with low identity, you can descrease word_size from 10 to minimum or so (e.g., -word_size 4 or -word_size 6) and increase e-value from 10 to some big value (e.g., -evalue 1000). You can also turn off masking of low-complexity sequence regions (-dust no -soft_masking false).

blastn -task blastn -query query.fasta -db db.fasta -word_size 4 -evalue 1000 -dust no -soft_masking false

If you are looking for very short DNA sequences you can add the -task blastn-short parameter and play with scoring parameters.

blastn -task blastn-short -query query.fasta -db db.fasta -word_size 4 -evalue 1000 -dust no -soft_masking false -gapopen 3 -penalty -1

[Joe]

You should add the -task blastn parameter (for finding sequences with low identity) even though your DNA sequence are longer than 50 bases. Because the default parameter is megablast, which is used to find very similar sequences.

[Andrzej Zielezinski]

Good point, I forgot about this. I updated my answer. Thanks.