Hi Joern,

thanks for your email. You were right : I have extensively explored the
bowtie manual, and I have been doing most of the analyses by using
-- best option. My apologies to the bioc-sig subscribers 'cause my email is
not 100% related to R/ShortRead, although I would appreciate
any comments on any of these 2 questions :

-  on a statistical ground, if read1 aligns to 5 places, any place is
selected with 0.2 probability. and if a read 2 aligns to 3 places, any place
is
chosen with 0.3 probability. therefore, having read1 and read2 in the same
place X would have the probability 0.3 * 0.2 . and if read3 ... read 7
are also included, the probability would be 0.3 * 0.2. * prob(read3) * ... *
probab(read7). if this is the case, would it be legitimate to assign a
ChIP-seq peak to the place X ?

- if bowtie outputs the reads that align to repetitive regions (and
arbitrarily selects one of these regions), could it be a good approximation
for
PET-seq data (that is intended to map the peaks on these repetitive regions,
but it is much more expensive) ?

thanks very much,

bogdan


On Tue, Jul 7, 2009 at 1:46 AM, Joern Toedling <[email protected]>wrote:

> Hello,
>
> I think you should better ask questions of this kind to the developers of
> Bowtie rather than on this mailing list. However, I believe all of your
> questions are answered in the manual of bowtie. Basically you can customize
> the report output by supplying additional arguments to the bowtie call.
> "-k"
> specifies the maximum number of valid match postions for each read in the
> output. If you set "-k 1" (the default) I think a random position is
> returned.
> To my mind, the score is not directly given but the output column gives you
> details which bases do not match and since you also have the quality
> information of each base you can easily calculate the sum of mismatching
> base
> qualities.
> The tab-delimited output format (non-binary) can be read in by function
> readAligned(..., type="bowtie")
> and the resulting object of class AlignedRead also stores the mismatch
> information (at least in the current development version of package
> ShortRead).
>
> Best regards,
> Joern
>
> On Mon, 6 Jul 2009 14:17:34 -0700, Bogdan Tanasa wrote
> > Hi everyone.
> >
> > I would appreciate to have your comments on the following : when aligning
> > the solexa reads with bowtie,
> > if a read aligns to multiple genomic regions, is the highest-scored
> location
> > picked up in the final report
> >
> > (i.e. when using --best option) ? And if a read aligns with the same
> > score to multiple regions, would it be possible to see the score of
> > the alignment and the differences in the score among multiple
> > regions ? In this last scenario, a randomly picked location among
> > the equally scored genomic locations is reported ?
> >
> > thanks very much,
> >
> > bogdan
>
> ---
> Joern Toedling
> Institut Curie -- U900
> 26 rue d'Ulm, 75005 Paris, FRANCE
> Tel. +33 (0)156246926
>
>

        [[alternative HTML version deleted]]

_______________________________________________
Bioc-sig-sequencing mailing list
[email protected]
https://stat.ethz.ch/mailman/listinfo/bioc-sig-sequencing

Reply via email to