Hi all,
I am new to the field of seq and performed a RIP-Seq experiment using
HTSeq count as counter.
I get now the following (using union, but doesn??t look better for
interesection_strict):
__no_feature 1503377
__ambiguous 490772
__too_low_aQual 0
__not_aligned 0
__alignment_not_unique 5277314
When I sum up counts for all genes, I get 3227845.
The number for __no_feature, __ambiguous, __alignment_not_unique look
very high.
Does somebody have an idea for that?
(Additional info: We did random priming and mapped with STAR and
masked rRNA loci)
Best wishes
Julia
-- output of sessionInfo():
.
--
Sent via the guest posting facility at bioconductor.org.
Hi Julia,
On Fri, Sep 5, 2014 at 7:41 AM, Julia [guest] <guest at="" bioconductor.org=""> wrote:
> Hi all,
> I am new to the field of seq and performed a RIP-Seq experiment
using HTSeq count as counter.
> I get now the following (using union, but doesn??t look better for
interesection_strict):
> __no_feature 1503377
> __ambiguous 490772
> __too_low_aQual 0
> __not_aligned 0
> __alignment_not_unique 5277314
>
> When I sum up counts for all genes, I get 3227845.
>
> The number for __no_feature, __ambiguous, __alignment_not_unique
look very high.
>
> Does somebody have an idea for that?
While I haven't worked with RIP-seq data myself, I do have some
experience with HITS-CLIP and PAR-CLIP, which I believe are quite
similar (at least in principle) -- these experiments must incredibly
difficult to pull off, however, because I'd say most of these types of
datasets that came my way were notoriously/incredibly noisy.
This is just to say the problem you are seeing may not be an
informatics problem, and could (quite possibly) be an experimental
one.
-steve
--
Steve Lianoglou
Computational Biologist
Genentech
Hi Steve,
thank you very much for your help.
Which tool for DE did you use?
I used edger, however I?ve read that edgeR and DESeq2 might be
overstringent for RIP-Seq (f.e. RIP-Seeker package Paper, Supplement).
Best wishes,
Julia
-----Urspr?ngliche Nachricht-----
Von: mailinglist.honeypot at gmail.com [mailto:mailinglist.honeypot at
gmail.com] Im Auftrag von Steve Lianoglou
Gesendet: Freitag, 5. September 2014 18:35
An: Julia [guest]
Cc: bioconductor at r-project.org list; Pickl, Julia
Betreff: Re: [BioC] HTSeq-Count
Hi Julia,
On Fri, Sep 5, 2014 at 7:41 AM, Julia [guest] <guest at="" bioconductor.org=""> wrote:
> Hi all,
> I am new to the field of seq and performed a RIP-Seq experiment
using HTSeq count as counter.
> I get now the following (using union, but doesn??t look better for
interesection_strict):
> __no_feature 1503377
> __ambiguous 490772
> __too_low_aQual 0
> __not_aligned 0
> __alignment_not_unique 5277314
>
> When I sum up counts for all genes, I get 3227845.
>
> The number for __no_feature, __ambiguous, __alignment_not_unique
look very high.
>
> Does somebody have an idea for that?
While I haven't worked with RIP-seq data myself, I do have some
experience with HITS-CLIP and PAR-CLIP, which I believe are quite
similar (at least in principle) -- these experiments must incredibly
difficult to pull off, however, because I'd say most of these types of
datasets that came my way were notoriously/incredibly noisy.
This is just to say the problem you are seeing may not be an
informatics problem, and could (quite possibly) be an experimental
one.
-steve
--
Steve Lianoglou
Computational Biologist
Genentech
Hi Julia,
On Mon, Sep 8, 2014 at 10:10 AM, Pickl, Julia
<j.pickl at="" dkfz-heidelberg.de=""> wrote:
> Hi Steve,
>
> thank you very much for your help.
>
> Which tool for DE did you use?
> I used edger, however I?ve read that edgeR and DESeq2 might be
overstringent for RIP-Seq (f.e. RIP-Seeker package Paper, Supplement).
At the time, we used DESeq. I was comparing differential binding of an
RBP to targets between conditions, though. It seems like you want to
look at one condition and ask what transcripts are being bound by a
RBP, though.
In my opinion, the differential binding question is more interesting
and biologically relevant -- and likely to give you "real" signal.
Again it's my opinion, but doing one experiment and asking where your
RBP binds in that experiment in isolation (even though you have done a
control run in the same cell type/condition) likely won't mean all
that much.
-steve
--
Steve Lianoglou
Computational Biologist
Genentech
Hi Julia,
You obviously allowed multiple mappings when calling STAR, however,
HTSeq counts only uniquely mapped reads. ~1.5M reads mapped to
intergenic and/or intronic regions which contributed to ?_no_feature?.
?_ambiguous? mapped to regions annotated by multiple genes.
If you want to take into account multiple mapped reads, you can either
evenly distribute them to all gene targets (normalized by times of
mapping), distribute them according to uniquely mapped reads, or
distribute them more sophistically by doing a multiple-run EM
algorithm (such as the RSEM does).
Cheers,
Yuan
On Sep 5, 2014, at 10:41 AM, Julia [guest] <guest at="" bioconductor.org="">
wrote:
> Hi all,
> I am new to the field of seq and performed a RIP-Seq experiment
using HTSeq count as counter.
> I get now the following (using union, but doesn??t look better for
interesection_strict):
> __no_feature 1503377
> __ambiguous 490772
> __too_low_aQual 0
> __not_aligned 0
> __alignment_not_unique 5277314
>
> When I sum up counts for all genes, I get 3227845.
>
> The number for __no_feature, __ambiguous, __alignment_not_unique
look very high.
>
> Does somebody have an idea for that?
>
> (Additional info: We did random priming and mapped with STAR and
masked rRNA loci)
>
> Best wishes
> Julia
>
> -- output of sessionInfo():
>
> .
>
> --
> Sent via the guest posting facility at bioconductor.org.
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at r-project.org
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives:
http://news.gmane.org/gmane.science.biology.informatics.conductor
[[alternative HTML version deleted]]
Hi Yuan,
thank you very much for your reply.
Would you say that I should take into account multiple mapped reads as
these are more than unique reads or do you think the high number of
counts for _alignment_not_unique are not a problem per se?
I am a beginner to this field, so I would be happy if you could share
your experience with me.
Best wishes
Julia
Von: Yuan Hao [mailto:yuan.x.hao at gmail.com]
Gesendet: Freitag, 5. September 2014 17:18
An: Julia [guest]
Cc: bioconductor at r-project.org; Pickl, Julia
Betreff: Re: [BioC] HTSeq-Count
Hi Julia,
You obviously allowed multiple mappings when calling STAR, however,
HTSeq counts only uniquely mapped reads. ~1.5M reads mapped to
intergenic and/or intronic regions which contributed to "_no_feature".
"_ambiguous" mapped to regions annotated by multiple genes.
If you want to take into account multiple mapped reads, you can either
evenly distribute them to all gene targets (normalized by times of
mapping), distribute them according to uniquely mapped reads, or
distribute them more sophistically by doing a multiple-run EM
algorithm (such as the RSEM does).
Cheers,
Yuan
On Sep 5, 2014, at 10:41 AM, Julia [guest] <guest at="" bioconductor.org<mailto:guest="" at="" bioconductor.org="">> wrote:
Hi all,
I am new to the field of seq and performed a RIP-Seq experiment using
HTSeq count as counter.
I get now the following (using union, but doesn?t look better for
interesection_strict):
__no_feature 1503377
__ambiguous 490772
__too_low_aQual 0
__not_aligned 0
__alignment_not_unique 5277314
When I sum up counts for all genes, I get 3227845.
The number for __no_feature, __ambiguous, __alignment_not_unique look
very high.
Does somebody have an idea for that?
(Additional info: We did random priming and mapped with STAR and
masked rRNA loci)
Best wishes
Julia
-- output of sessionInfo():
.
--
Sent via the guest posting facility at
bioconductor.org<http: bioconductor.org="">.
_______________________________________________
Bioconductor mailing list
Bioconductor at r-project.org<mailto:bioconductor at="" r-project.org="">
https://stat.ethz.ch/mailman/listinfo/bioconductor
Search the archives:
http://news.gmane.org/gmane.science.biology.informatics.conductor
[[alternative HTML version deleted]]
Hi Julia,
Depending on questions in hand, there are good reasons to consider
only uniquely mappable reads or being serious about multiple mappings
(such as studying TEs or facing repetitive genomes). In terms of
traditional RNASeq data, however, most time I ran into situation that
multiple reads contributed to a substential part and personally I
believe including the multiple reads should benefit the accurate
expression estimation.
Cheers,
Yuan
On Sep 5, 2014, at 11:48 AM, Pickl, Julia <j.pickl at="" dkfz-="" heidelberg.de=""> wrote:
> Hi Yuan,
>
> thank you very much for your reply.
>
> Would you say that I should take into account multiple mapped reads
as these are more than unique reads or do you think the high number
of counts for _alignment_not_unique are not a problem per se?
>
> I am a beginner to this field, so I would be happy if you could
share your experience with me.
>
> Best wishes
> Julia
>
> Von: Yuan Hao [mailto:yuan.x.hao at gmail.com]
> Gesendet: Freitag, 5. September 2014 17:18
> An: Julia [guest]
> Cc: bioconductor at r-project.org; Pickl, Julia
> Betreff: Re: [BioC] HTSeq-Count
>
> Hi Julia,
>
> You obviously allowed multiple mappings when calling STAR, however,
HTSeq counts only uniquely mapped reads. ~1.5M reads mapped to
intergenic and/or intronic regions which contributed to ?_no_feature?.
?_ambiguous? mapped to regions annotated by multiple genes.
>
> If you want to take into account multiple mapped reads, you can
either evenly distribute them to all gene targets (normalized by times
of mapping), distribute them according to uniquely mapped reads, or
distribute them more sophistically by doing a multiple-run EM
algorithm (such as the RSEM does).
>
> Cheers,
> Yuan
> On Sep 5, 2014, at 10:41 AM, Julia [guest] <guest at="" bioconductor.org=""> wrote:
>
>
> Hi all,
> I am new to the field of seq and performed a RIP-Seq experiment
using HTSeq count as counter.
> I get now the following (using union, but doesn??t look better for
interesection_strict):
> __no_feature 1503377
> __ambiguous 490772
> __too_low_aQual 0
> __not_aligned 0
> __alignment_not_unique 5277314
>
> When I sum up counts for all genes, I get 3227845.
>
> The number for __no_feature, __ambiguous, __alignment_not_unique
look very high.
>
> Does somebody have an idea for that?
>
> (Additional info: We did random priming and mapped with STAR and
masked rRNA loci)
>
> Best wishes
> Julia
>
> -- output of sessionInfo():
>
> .
>
> --
> Sent via the guest posting facility at bioconductor.org.
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at r-project.org
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives:
http://news.gmane.org/gmane.science.biology.informatics.conductor
[[alternative HTML version deleted]]
Hi Yuan,
thank you very much for sharing your experience with me!
Best wishes
Julia
Von: Yuan Hao [mailto:yuan.x.hao at gmail.com]
Gesendet: Freitag, 5. September 2014 19:45
An: Pickl, Julia
Cc: Yuan Hao; Julia [guest]; bioconductor at r-project.org
Betreff: Re: [BioC] HTSeq-Count
Hi Julia,
Depending on questions in hand, there are good reasons to consider
only uniquely mappable reads or being serious about multiple mappings
(such as studying TEs or facing repetitive genomes). In terms of
traditional RNASeq data, however, most time I ran into situation that
multiple reads contributed to a substential part and personally I
believe including the multiple reads should benefit the accurate
expression estimation.
Cheers,
Yuan
On Sep 5, 2014, at 11:48 AM, Pickl, Julia <j.pickl at="" dkfz-="" heidelberg.de<mailto:j.pickl="" at="" dkfz-heidelberg.de="">> wrote:
Hi Yuan,
thank you very much for your reply.
Would you say that I should take into account multiple mapped reads as
these are more than unique reads or do you think the high number of
counts for _alignment_not_unique are not a problem per se?
I am a beginner to this field, so I would be happy if you could share
your experience with me.
Best wishes
Julia
Von: Yuan Hao [mailto:yuan.x.hao at gmail.com]
Gesendet: Freitag, 5. September 2014 17:18
An: Julia [guest]
Cc: bioconductor at r-project.org<mailto:bioconductor at="" r-project.org="">; Pickl, Julia
Betreff: Re: [BioC] HTSeq-Count
Hi Julia,
You obviously allowed multiple mappings when calling STAR, however,
HTSeq counts only uniquely mapped reads. ~1.5M reads mapped to
intergenic and/or intronic regions which contributed to "_no_feature".
"_ambiguous" mapped to regions annotated by multiple genes.
If you want to take into account multiple mapped reads, you can either
evenly distribute them to all gene targets (normalized by times of
mapping), distribute them according to uniquely mapped reads, or
distribute them more sophistically by doing a multiple-run EM
algorithm (such as the RSEM does).
Cheers,
Yuan
On Sep 5, 2014, at 10:41 AM, Julia [guest] <guest at="" bioconductor.org<mailto:guest="" at="" bioconductor.org="">> wrote:
Hi all,
I am new to the field of seq and performed a RIP-Seq experiment using
HTSeq count as counter.
I get now the following (using union, but doesn?t look better for
interesection_strict):
__no_feature 1503377
__ambiguous 490772
__too_low_aQual 0
__not_aligned 0
__alignment_not_unique 5277314
When I sum up counts for all genes, I get 3227845.
The number for __no_feature, __ambiguous, __alignment_not_unique look
very high.
Does somebody have an idea for that?
(Additional info: We did random priming and mapped with STAR and
masked rRNA loci)
Best wishes
Julia
-- output of sessionInfo():
.
--
Sent via the guest posting facility at
bioconductor.org<http: bioconductor.org=""/>.
_______________________________________________
Bioconductor mailing list
Bioconductor at r-project.org<mailto:bioconductor at="" r-project.org="">
https://stat.ethz.ch/mailman/listinfo/bioconductor
Search the archives:
http://news.gmane.org/gmane.science.biology.informatics.conductor
[[alternative HTML version deleted]]
