Does BLAST -max_target_seqs Number of aligned sequences to keep option influence total running time ?

Question

[PRog]

Hi,

Navigating across posts related to BLAST -max_target_seqs option or about limiting the number aligned sequences to keep, I could not find any definitive answer about whether or not this option can significantly affect the total running time ?

Of course it affects the writing time, in that reporting 5 hits is quicker than reporting a million but, in the case of BLASTs against the nr db for example, the writing time seems often negligible in comparison with the screening-of-the-db-and-alignment time.

So my main question is at the end : Does BLAST stops when "x" sequences are aligned (with -max_target_seqs x)?

Trying with a few sequences for a reasonable test time let me circumspect whether or not it makes a difference.

Answer 1

[GenoMax]

There is a response from NCBI about this specific parameter : https://academic.oup.com/bioinformatics/article/35/15/2699/5259186

Bottom line is NCBI's recommendation is that you should use 5 for that option and then filter afterwards. Best hit may not be the top one. If you are not worried about this being the case then by all means use 1.

[PRog]

From the response : "BLAST does process every sequence in its search set. It does not, as Shah et al. (2018) state, simply return the ‘first N hits that exceed the specified E-value threshold’[...]". So I suppose its is bright clear : whatever '-max_target_seqs' number, BLAST always screens the entiere db. Thks !