I wonder if anyone has encountered this problem: For < 80 files,
Peptideprophet works normally, but when there are more than 80 files,
it doesn't start at all.
If we split them up into 2 batches to run peptideprophet, the
statistical distributions that we obtain would be different since it
may happen that we unknowingly choose a group of poor scoring spectra
to model for 1 batch, while the other batch contains high scoring
ones. Would the peptideprophet probabilities obtained for the poor
scoring batch be reliable ? Ideally we should analyse the distribution
across all the files from a single sample (separated into different
protein fractions) altogether, isn't it?
Would really appreciate if someone could enlighten on this, and
whether this can be gotten around by simply merging the some of the
smaller pepxml files in the first place.
What operating system are you on (windows, linux etc)? My guess is that
it's probably about the length of the command line, rather than the file
count. We've had issues with this in the past, it might be borken again.
[mailto:spctools-discuss@googlegroups.com] On Behalf Of Bernt
Sent: Wednesday, July 01, 2009 3:14 AM
To: spctools-discuss
Subject: [spctools-discuss] Does Peptideprophet accept only a limited no. of
pepXML files?
Dear all,
I wonder if anyone has encountered this problem: For < 80 files,
Peptideprophet works normally, but when there are more than 80 files,
it doesn't start at all.
If we split them up into 2 batches to run peptideprophet, the
statistical distributions that we obtain would be different since it
may happen that we unknowingly choose a group of poor scoring spectra
to model for 1 batch, while the other batch contains high scoring
ones. Would the peptideprophet probabilities obtained for the poor
scoring batch be reliable ? Ideally we should analyse the distribution
across all the files from a single sample (separated into different
protein fractions) altogether, isn't it?
Would really appreciate if someone could enlighten on this, and
whether this can be gotten around by simply merging the some of the
smaller pepxml files in the first place.
Thanks for your response. I'm using TPPv4.2 Rev 1 on Windows XP. I
read the thread on "limit on number of pep.xml files that can be
analysed with interact" before, and Eric Deutsch was suggesting using
<xinteract *.pep.xml> on the command line. I tried it, and it started
processing all the 100 over files i had, but in the end there was a
problem with the generated interact.pep.xml file. Were you going to
suggest this <xinteract *.pep.xml> method?
Bernt
On Jul 1, 11:32 pm, "Brian Pratt" <brian.pr...@insilicos.com> wrote:
> What operating system are you on (windows, linux etc)? My guess is that
> it's probably about the length of the command line, rather than the file
> count. We've had issues with this in the past, it might be borken again.
> [mailto:spctools-discuss@googlegroups.com] On Behalf Of Bernt
> Sent: Wednesday, July 01, 2009 3:14 AM
> To: spctools-discuss
> Subject: [spctools-discuss] Does Peptideprophet accept only a limited no. of
> pepXML files?
> Dear all,
> I wonder if anyone has encountered this problem: For < 80 files,
> Peptideprophet works normally, but when there are more than 80 files,
> it doesn't start at all.
> If we split them up into 2 batches to run peptideprophet, the
> statistical distributions that we obtain would be different since it
> may happen that we unknowingly choose a group of poor scoring spectra
> to model for 1 batch, while the other batch contains high scoring
> ones. Would the peptideprophet probabilities obtained for the poor
> scoring batch be reliable ? Ideally we should analyse the distribution
> across all the files from a single sample (separated into different
> protein fractions) altogether, isn't it?
> Would really appreciate if someone could enlighten on this, and
> whether this can be gotten around by simply merging the some of the
> smaller pepxml files in the first place.
[mailto:spctools-discuss@googlegroups.com] On Behalf Of Bernt
Sent: Thursday, July 02, 2009 10:15 AM
To: spctools-discuss
Subject: [spctools-discuss] Re: Does Peptideprophet accept only a limited
no. of pepXML files?
Hi Brian,
Thanks for your response. I'm using TPPv4.2 Rev 1 on Windows XP. I
read the thread on "limit on number of pep.xml files that can be
analysed with interact" before, and Eric Deutsch was suggesting using
<xinteract *.pep.xml> on the command line. I tried it, and it started
processing all the 100 over files i had, but in the end there was a
problem with the generated interact.pep.xml file. Were you going to
suggest this <xinteract *.pep.xml> method?
Bernt
On Jul 1, 11:32 pm, "Brian Pratt" <brian.pr...@insilicos.com> wrote:
> What operating system are you on (windows, linux etc)? My guess is that
> it's probably about the length of the command line, rather than the file
> count. We've had issues with this in the past, it might be borken again.
> [mailto:spctools-discuss@googlegroups.com] On Behalf Of Bernt
> Sent: Wednesday, July 01, 2009 3:14 AM
> To: spctools-discuss
> Subject: [spctools-discuss] Does Peptideprophet accept only a limited no.
of
> pepXML files?
> Dear all,
> I wonder if anyone has encountered this problem: For < 80 files,
> Peptideprophet works normally, but when there are more than 80 files,
> it doesn't start at all.
> If we split them up into 2 batches to run peptideprophet, the
> statistical distributions that we obtain would be different since it
> may happen that we unknowingly choose a group of poor scoring spectra
> to model for 1 batch, while the other batch contains high scoring
> ones. Would the peptideprophet probabilities obtained for the poor
> scoring batch be reliable ? Ideally we should analyse the distribution
> across all the files from a single sample (separated into different
> protein fractions) altogether, isn't it?
> Would really appreciate if someone could enlighten on this, and
> whether this can be gotten around by simply merging the some of the
> smaller pepxml files in the first place.
What was the problem with the final generated interact.pep.xml file?
Can you verify that all of your pep.xml files are complete? Check
their size and starting and ending tags for completeness. If there
are any error message please post them here.
On Thu, Jul 2, 2009 at 10:40 AM, Brian Pratt<brian.pr...@insilicos.com> wrote:
> That's the issue I was thinking about, yes - so it sounds like using the
> "xinteract *.pep.xm" approach you're on to a new and different problem,
> then?
> -----Original Message-----
> From: spctools-discuss@googlegroups.com
> [mailto:spctools-discuss@googlegroups.com] On Behalf Of Bernt
> Sent: Thursday, July 02, 2009 10:15 AM
> To: spctools-discuss
> Subject: [spctools-discuss] Re: Does Peptideprophet accept only a limited
> no. of pepXML files?
> Hi Brian,
> Thanks for your response. I'm using TPPv4.2 Rev 1 on Windows XP. I
> read the thread on "limit on number of pep.xml files that can be
> analysed with interact" before, and Eric Deutsch was suggesting using
> <xinteract *.pep.xml> on the command line. I tried it, and it started
> processing all the 100 over files i had, but in the end there was a
> problem with the generated interact.pep.xml file. Were you going to
> suggest this <xinteract *.pep.xml> method?
> Bernt
> On Jul 1, 11:32 pm, "Brian Pratt" <brian.pr...@insilicos.com> wrote:
>> What operating system are you on (windows, linux etc)? My guess is that
>> it's probably about the length of the command line, rather than the file
>> count. We've had issues with this in the past, it might be borken again.
>> [mailto:spctools-discuss@googlegroups.com] On Behalf Of Bernt
>> Sent: Wednesday, July 01, 2009 3:14 AM
>> To: spctools-discuss
>> Subject: [spctools-discuss] Does Peptideprophet accept only a limited no.
> of
>> pepXML files?
>> Dear all,
>> I wonder if anyone has encountered this problem: For < 80 files,
>> Peptideprophet works normally, but when there are more than 80 files,
>> it doesn't start at all.
>> If we split them up into 2 batches to run peptideprophet, the
>> statistical distributions that we obtain would be different since it
>> may happen that we unknowingly choose a group of poor scoring spectra
>> to model for 1 batch, while the other batch contains high scoring
>> ones. Would the peptideprophet probabilities obtained for the poor
>> scoring batch be reliable ? Ideally we should analyse the distribution
>> across all the files from a single sample (separated into different
>> protein fractions) altogether, isn't it?
>> Would really appreciate if someone could enlighten on this, and
>> whether this can be gotten around by simply merging the some of the
>> smaller pepxml files in the first place.
Thanks for your response, I realised the xinteract has to be used with
*.pep.xml behind the options. Previously I used it before the options.
It manages to generate a working interact.pep.xml with all the
PeptideProphet probabilities. But the .shtml file showed an error:
This is not a big problem for me. I can live with it. The problem that
remains is the ion spectra could not be viewed in the PepXML browser.
I have put it in a separate post, since it should not be due to the
reason of combining >100 pep.xml files on the command line. Thanks for
your help too, Brian.
Bernard
On Jul 3, 2:24 am, David Shteynberg <dshteynb...@systemsbiology.org>
wrote:
> What was the problem with the final generated interact.pep.xml file?
> Can you verify that all of your pep.xml files are complete? Check
> their size and starting and ending tags for completeness. If there
> are any error message please post them here.
> -David
> On Thu, Jul 2, 2009 at 10:40 AM, Brian Pratt<brian.pr...@insilicos.com> wrote:
> > That's the issue I was thinking about, yes - so it sounds like using the
> > "xinteract *.pep.xm" approach you're on to a new and different problem,
> > then?
> > -----Original Message-----
> > From: spctools-discuss@googlegroups.com
> > [mailto:spctools-discuss@googlegroups.com] On Behalf Of Bernt
> > Sent: Thursday, July 02, 2009 10:15 AM
> > To: spctools-discuss
> > Subject: [spctools-discuss] Re: Does Peptideprophet accept only a limited
> > no. of pepXML files?
> > Hi Brian,
> > Thanks for your response. I'm using TPPv4.2 Rev 1 on Windows XP. I
> > read the thread on "limit on number of pep.xml files that can be
> > analysed with interact" before, and Eric Deutsch was suggesting using
> > <xinteract *.pep.xml> on the command line. I tried it, and it started
> > processing all the 100 over files i had, but in the end there was a
> > problem with the generated interact.pep.xml file. Were you going to
> > suggest this <xinteract *.pep.xml> method?
> > Bernt
> > On Jul 1, 11:32 pm, "Brian Pratt" <brian.pr...@insilicos.com> wrote:
> >> What operating system are you on (windows, linux etc)? My guess is that
> >> it's probably about the length of the command line, rather than the file
> >> count. We've had issues with this in the past, it might be borken again.
> >> [mailto:spctools-discuss@googlegroups.com] On Behalf Of Bernt
> >> Sent: Wednesday, July 01, 2009 3:14 AM
> >> To: spctools-discuss
> >> Subject: [spctools-discuss] Does Peptideprophet accept only a limited no.
> > of
> >> pepXML files?
> >> Dear all,
> >> I wonder if anyone has encountered this problem: For < 80 files,
> >> Peptideprophet works normally, but when there are more than 80 files,
> >> it doesn't start at all.
> >> If we split them up into 2 batches to run peptideprophet, the
> >> statistical distributions that we obtain would be different since it
> >> may happen that we unknowingly choose a group of poor scoring spectra
> >> to model for 1 batch, while the other batch contains high scoring
> >> ones. Would the peptideprophet probabilities obtained for the poor
> >> scoring batch be reliable ? Ideally we should analyse the distribution
> >> across all the files from a single sample (separated into different
> >> protein fractions) altogether, isn't it?
> >> Would really appreciate if someone could enlighten on this, and
> >> whether this can be gotten around by simply merging the some of the
> >> smaller pepxml files in the first place.
|
