4
$\begingroup$

After some unexpected results (and previously reported) I heard that there are other tools for finding similar sequences besides blast that are faster and more accurate.

I only found hmmer, but I don't know if it is faster and more accurate.

Is there any review comparing such tools? A quick google scholar search didn't report anything useful.

I want to search with short DNA sequences (~400), mainly interested for taxonomic imputation so I'll be looking for similar sequences and not distant homologs. I won't try to model genes and the sequence will be unspliced as it is from direct sequencing of a a region of the 16S gene.

$\endgroup$
  • $\begingroup$ Please edit your question and give us more details about what you will be looking for. Will these be nucleotide queries? Protein? Will you be looking for very similar sequences or distant homologs? Long sequences? Short ones? These details will affect your choice of tool. $\endgroup$ – terdon Nov 26 '18 at 12:11
  • $\begingroup$ My question is mainly theoretically, but I narrowed with the use case that prompted me to ask/think about this $\endgroup$ – llrs Nov 26 '18 at 12:56
  • 1
    $\begingroup$ The thing is you have tools like blat and megablat, FASTA (although I doubt anyone uses that anymore), hmmer, cd-hit, psi-blast, etc. It's really hard to answer unless you give a specific usage scenario. For instance, will you be trying to model genes? Would you want a tool like exonerate or genewise that can align a spliced cDNA sequence to a genome? Or will your 400 sequences be unspliced? $\endgroup$ – terdon Nov 26 '18 at 13:12
  • $\begingroup$ Ok, I realize this is more complicated than I thought. Thanks it is not my field/research focus :D. I updated the question with more information. Thanks for the guidance. $\endgroup$ – llrs Nov 26 '18 at 13:24
  • $\begingroup$ Try VSEARCH (or USEARCH). They are not BLAST replacement but can be a good choice for the right applications. $\endgroup$ – user172818 Nov 27 '18 at 1:10
4
$\begingroup$

As I understand, the software tool Lambda is a viable, yet lesser known alternative to BLAST in the context of taxonomic classification of NGS data.

| improve this answer | |
$\endgroup$
4
$\begingroup$

Besides generic nucleotide aligners, there are also more specialized tools for the alignment of 16S amplicon sequences, e.g. SINA (article, software) which is part of SILVA

| improve this answer | |
$\endgroup$
0
$\begingroup$

Why not shift the parameter inputs of Blast to exclude the species/subspecies which are being queried? Happy to show you how to do it, but that will pick up homologues. Alternative parse the blast output for e.g. 20000 outputs using E or identity within the threshold required (that's the way I'd do it).

| improve this answer | |
$\endgroup$
0
$\begingroup$

All tools I’m aware of will be more-or-less as fast as blast. With kmer-based methods, the investment will be the up-front kmer database build.

Not sure if you’ve tried this, but NCBI has a 16S microbial blast database (lots of species, 16S rRNA only) just for this purpose. No point to search millions of nts for the 16S sequence every time if you know it’s SOME 16S sequence, so limit the search to only16S sequences.

I’d recommend you build a local blast database from ftp://ftp.ncbi.nih.gov/blast/db/16SMicrobial.tar.gz

BTW, another answer says blast won’t give you the alignment. Blast actually will give you alignments, you just have to ask it for that output format.

| improve this answer | |
$\endgroup$
-1
$\begingroup$

You can give a try to this tool: dnaservic.es It's actually available experimentaly. It is a real local aligner, not just a search tool like BLAST. It really returns an alignment. It is usually slower than BLAST, but bot much slower in most cases. In contrast it usually returns alignments with better scores and not only partial alignments.

| improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.