SEQanswers

SEQanswers (http://seqanswers.com/forums/index.php)
-   Bioinformatics (http://seqanswers.com/forums/forumdisplay.php?f=18)
-   -   A question about tophat -g/--max-multihits option (http://seqanswers.com/forums/showthread.php?t=8548)

keycard 12-21-2010 06:35 PM

A question about tophat -g/--max-multihits option
 
If the --max-multihits option of tophat is set to 1, am I going to get the "best" alignment for the read? For example, will tophat pick the alignment for a read based on the best mapping score or the minimun number of mismatches? I am wondering how tophat works if there are multiple hits and the --max-multihits option is set to 1. Thanks.

brentp 12-22-2010 07:02 AM

i think you need to use -g , not --max-multihits due to a bug in the arg parsing in tophat.
but if you use -g 1, then you will get the best alignment and send -m 1 to bowtie.
this: http://bowtie-bio.sourceforge.net/ma...e-options-best
describes how the "best" is determined. it looks at mismatches, then quality.

Boel 12-29-2010 04:36 AM

I don't believe that is correct that setting --max-multihits to 1 gives you the BEST alignment for the read. According to the tophat manual --max-multihits 1 will trow away all reads that have more than 1 valid alignment. Hence you will get back sam files that only contain the reads that have exactly one alignment if this option is set to 1.

Quote:

Instructs TopHat to allow up to this many alignments to the reference for a given read, and suppresses all alignments for reads with more than this many alignments. The default is 40.


All times are GMT -8. The time now is 01:48 AM.

Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2021, vBulletin Solutions, Inc.