An Alternative To Blastx With Speed Similar To Bowtie

Question

[Tianyang Li]

Hi,

Is there any fast aligner that aligns nucleotide sequences to protein databases?

I've used blastx but blastx is too slow and takes too much memory, I'm wondering if anything like bowtie exists.

Thanks!

Answer 1

[Martin A Hansen]

Maybe Usearch might be useful.

Answer 2

[Neilfws]

You could try BLAT (downloads) with the options:

-t=prot -q=dnax

I've only used it for nucleotide-nucleotide searches, but it should be faster than BLAST. You just need sufficient RAM for the in-memory database indices.

[Tianyang Li]

Apparently blat can't do this. I got this when I ran it: d and q must both be either protein or dna

[Neilfws]

Really? Interesting, I've never tried and just assumed it had a blastx-like function.

[Neilfws]

A new answer brought me back to this old thread.

The error message "d and q must both be either protein or dna" is not correct; it should begin "t and q". Furthermore, this error occurs when -q=dna, not -q=dnax as I suggested. The error in the latter case is "Unrecognized combination of target and query types."

An answer in this thread suggests simply switching the -t and -q options (i.e. -q=prot -t=dnax).

[Prakki Rama]

Blat with options -t=prot -q=dnax gives results similar to tBLASTn but not BLASTx. Be careful during the comparisons.

Answer 3

[Dgg32]

RAPsearch 2 can do DNA vs protein or protein vs protein, very fast.

Answer 4

[Len Trigg]

The mapx command from Real Time Genomics is for this specific scenario. As well as being inherently very fast, it also supports multiple cores. It is included in our commercial RTG Metagenomics and RTG Core products, as well as the non-commercial version of RTG Core.

Answer 5

[h.mon]

I know I am reviving a really old thread, but it is surprising no one has cited DIAMOND so far. It is not particularly light on resources, but it is incredibly fast.

[lieven.sterck]

All true, but unfortunately it stinks at doing blastX-like searches

[Carambakaracho]

really? besides protein-protein searches, I understood this was one of the principal objectives of the algorithm. What do you mean?