diamond icon indicating copy to clipboard operation
diamond copied to clipboard
verified publisher icon bbuchfink

Does Diamond support DNA sequence (that cannot be translated into protein) alignment?

Open simon19891216 opened this issue 2 years ago • 3 comments

Thank you for your excellent work! We know that the parameter 'blastx' can be used for DNA alignment. We have no idea if a DNA sequence can not be translated into protein sequence, can Diamond still be used for alignment? If possible, how should we proceed?

simon19891216 avatar Sep 27 '23 02:09 simon19891216

You can try to use diamond blastx to align any DNA sequence, if it's not protein coding you will not find anything.

bbuchfink avatar Oct 04 '23 10:10 bbuchfink

You can try to use diamond blastx to align any DNA sequence, if it's not protein coding you will not find anything.

I've been goofing around just to see if I can and I fed diamond non-coding regions to use as targets. I translate these regions all to frame 1, then I do the six fold translation on the query and compare against it. It's SURPRISINGLY accurate at finding them even though they aren't protein coding. Basically, as long as it's conserved it hasn't mattered so far if it's coding or non-coding.

I'm still in the early stages so I expect it to fizzle out at some point but I've gotten farther than I expected...

Edit: (The hypothesis behind this is that if a region is highly conserved one might expect it to translate to roughly the same thing anyway even if it's "non-coding". Exceptions certainly apply...)

kevinmoran1988 avatar Oct 20 '23 01:10 kevinmoran1988

Thank you. In fact, I hope that DIAMOND can add a function similar to blastn in BLAST+, I firmly believe that this can better provide help for scientific research. Thanks again and best wishes.

At 2023-10-20 09:37:58, "Kevin Moran" @.***> wrote:

You can try to use diamond blastx to align any DNA sequence, if it's not protein coding you will not find anything.

I've been goofing around just to see if I can, and I fed diamond non-coding regions as references. I translate these regions all to frame 1, then I do the six fold translation on the query and compare against it. It's SURPRISINGLY accurate at finding them even though they aren't protein coding.

I'm still in the early stages so I expect it to fizzle out at some point but I've gotten farther than I expected...

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

simon19891216 avatar Oct 20 '23 01:10 simon19891216