SEQanswers

SEQanswers (http://seqanswers.com/forums/index.php)
-   Bioinformatics (http://seqanswers.com/forums/forumdisplay.php?f=18)
-   -   MegaBLAST vs NUCmer (http://seqanswers.com/forums/showthread.php?t=47830)

benjaminversteeg 10-28-2014 05:34 AM

MegaBLAST vs NUCmer
 
Hi,

For a project I'm trying to find out which pairwise aligner is best for aligning whole (bacterial) genomes.

Important aspects for choosing the best aligner are speed and sensitivity. In this situation NUCmer and BLAST seem both great programs for making local alignments. To find out which suits best, I've ran some tests. I found out that:
- MegaBLAST is a bit more sensitive than NUCmer, because it finds more alignments, but also breaks off alignments more quickly.
- At the aspect of speed NUCmer is BLASTed away with a factor of 3 to 4. Both programs are very fast though.

My question remains, what advantages would one have when using NUCmer instead of BLAST?

Thank you

Sheree Yau 10-28-2014 06:27 AM

I think you answered your own question with your tests. NUCmer is faster than MEGABLAST, but perhaps less sensitive. So depending on your project, is speed more important than accuracy?

Another consideration might be whether you favour longer alignment lengths of lower identity, for example for detecting synteny. From my experience, I tend to prefer BLAST since the time is not limiting until LOTS comparisons and I find there's many tools out there that take BLAST outputs.

I also like using Artemis Comparison Tool (ACT) (http://www.sanger.ac.uk/resources/software/act/) for visualising BLAST alignments. Good luck!

GenoMax 10-28-2014 06:30 AM

Do include mauve in your list of programs.

Sheree Yau 10-28-2014 06:45 AM

Yes agreed! MAUVE is very nice for comparing multiple bacterial genomes.

benjaminversteeg 10-28-2014 08:19 AM

Thank you for your response. Sorry for the confusion, but with the sentence about the speed I meant that MegaBLAST is much faster. I also measured Mauve by the way.

Tested with two bacterial samples (+/- 4.2MB) with five runs. See below average durations and the sum of all coverages (can exceed 100% because some regions are aligned at multiple locations):

- NUCmer: 15.52 s / 98.27%
- Mauve: 82.40 s / 103.60%
- MegaBLAST: 4.86 s / 100.91%
- BLASTn: 33.02 s / 122.36%

Speed is important for this project, but not crucial. The thing is though that MegaBLAST is faster, but also more sensitive.. Which makes me wonder why people nowadays still choose NUCmer instead of MegaBLAST. :confused:


All times are GMT -8. The time now is 02:54 AM.

Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2019, vBulletin Solutions, Inc.