Question: Cuffdiff secondary alignments (lower counts?)
0
gravatar for jomaco
5.3 years ago by
jomaco190
European Union
jomaco190 wrote:

Hi,

When I use the option --report-secondary-alignments in TopHat, cuffdiff reports that there are fewer fragments mapping to some genes in the genes.count_tracking file.

Without any filtering following TopHat I would expect counts to be increased in most cases when using the -report-secondary-alignments option? Does anyone know what might be going on here?

(My ultimate aim is to remove non-unique alignments, but when I do this (using samtools view -bq 4) without first reporting secondary alignments there appears to be no change in the results, so I started exploring other options i.e. --report-secondary-alignments).

Thanks

ADD COMMENTlink modified 4.2 years ago by bodington0 • written 5.3 years ago by jomaco190
1
gravatar for Devon Ryan
5.2 years ago by
Devon Ryan95k
Freiburg, Germany
Devon Ryan95k wrote:

This isn't really surprising. Tophat is likely just randomly producing a primary alignment at random. Given the expectation maximization that cufflinks is doing, some of these will end up getting reassigned when there are alternate (i.e., secondary) alignments available. So, some counts will go up, others down.

If you want only unique alignments the just do samtools view -q 4, since tophat assigns multimappers MAPQ values <= 3. BTW, if you're going to do that then it's questionable why you would want to bother with cufflinks/cuffdiff to begin with, at least if your genome has a decent annotation.

ADD COMMENTlink written 5.2 years ago by Devon Ryan95k

Thanks, that makes sense, which alternative tool would you recommend if that is the case (and why - presumably something to with it not being de novo)? 

ADD REPLYlink modified 5.1 years ago • written 5.1 years ago by jomaco190
0
gravatar for bodington
4.2 years ago by
bodington0
bodington0 wrote:

As far as I know, cufflinks actually splits the weighting of multiple alignments. If there are 10 alignments, each read will be assigned as 10% of a read at each alignment. Therefore, your counts will go down on some genes.

This explains it a lot better than I can.

ADD COMMENTlink written 4.2 years ago by bodington0
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1955 users visited in the last hour