iProphet Information

[bowe...@gmail.com]

I was wondering if someone could post some more detail on iProphet.
One thing that I would like to know is what all command line arguments
it takes, maybe some sample command lines, as that is how I work with
the TPP tools. I've gotten a little behind on some of the newer tools,
but I think iProphet is something I've been looking for.

Thanks,
Greg

[Greg Bowersock]

Just bumping this one up, since it looks like it got missed.

On Jun 10, 12:08 pm, "bowers...@gmail.com" <bowers...@gmail.com>
wrote:

[David Shteynberg]

Hi Greg,

The iProphet powerful tool that is very simple to run. Unless you are
testing the performance of the various models iProphet applies, there
is not need to set iProphet parameters. The easiest way to run the
tool is on a single analysis (one experiment, one search engine),
this only requires enabling the xinteract -ip option which will run
iProphet and ProteinProphet in IPROPHET mode in parallel to the
standard analysis.

A more complex analysis can be setup for combining multiple
experiments and multiple search engines. The individual search engines
and experiments must first be analyzed using xinteract (using option
-E<experiment_label>) where the different searches of the same data
get labeled with the same <experiment_label>. You can then run
iProphet on the PeptideProphet results using the InterProphetParser
tool.

Running it on the commandline without any arguments produces following
the usage statement:

usage: InterProphetParser <file1.pep.xml> <file2.pep.xml>... <outfile>

Where each of the input pep.xml files must come from a PeptideProphet
analysis, and the spectra that were searched with different search
engines have the same experiment_label spectrum_name combination. The
outfile is a name of the pep.xml file where iProphet will write the
results. For example:


> InterProphetParser exp1/sequest/interact.pep.xml exp1/xtandem/interact.pep.xml exp1/mascot/interact.pep.xml exp2/sequest/interact.pep.xml exp2/xtandem/interact.pep.xml exp2/mascot/interact.pep.xml interact.iproph.pep.xml

To run ProteinProphet on the iProphet results you can use the command

> ProteinProphet interact.iproph.pep.xml interact.iproph.prot.xml



-David

[David Shteynberg]

One mistake.... to run ProteinProphet on iProphet data you must use
the IPROPHET option as follows:

> ProteinProphet interact.iproph.pep.xml interact.iproph.prot.xml IPROPHET

[Greg Bowersock]

Thanks, I'll have to play around with it some to see how I like it. I do combine search engines, so it'll be interesting to compare the difference.

[Jake W]

David--

Thanks for your reply. Just to clarify: if one wants to use iProphet
to combine physically separate analyses (eg., multiple SCX fractions
or gel slices, with a separate MS/MS data file for each) from a single
biological 'experiment'/sample, should the -E label be used to
differentiate each fraction/slice/etc.? Or does the fact that each
scan bears a unique identifier from the MS data file name serve the
same function? Sorry if the question is off the mark, I just want to
be sure what a distinct "experiment" is in this context.

thank you,
Jake

[David Shteynberg]

Hi Jake,

It is up to you to define the experiment boundaries. I would imagine
that different fractions of SCX of a single sample, for instance, can
be classified as one experiment. As long as different fractions or
separate physical runs have different MS data file names there will be
no problem as far as iProphet having a name collision. If you are
combining several experiments (e.g. building an atlas of peptides from
different lab) then you should use a different experiment label for MS
runs from different experiments/labs, and when you use and experiment
label and are combining multiple search engines, you must use the same
experiment label for the all searches of the same runs.


-David

[Jake W]

David--

Thanks. That's how I've been using it and the results have looked
really good.

Jake

[Bernt]

Hi David,

With reference to the iProphet usage, as described in your post:

InterProphetParser exp1/sequest/interact.pep.xml exp1/xtandem/
interact.pep.xml exp1/mascot/interact.pep.xml exp2/sequest/
interact.pep.xml

Do the pepxml files for the different search results (i.e from mascot
and xtandem) have to be placed in separate folders to create that "/
xtandem/interact.pep.xml" file string recognition? If the
interact.pep.xml files are named differently to indicate the different
searches performed, is it not sufficient?

And do the pep.xml files for tandem and Mascot searches have to be
renamed with exactly the same extensions and filenames (i.e.
*.tandem.pep.xml -> *.pep.xml and *.dat.pep.xml -> *.pep.xml) for
PeptideProphet to add the -E expt label?


Bernt


On Jul 1, 2:39 am, David Shteynberg <dshteynb...@systemsbiology.org>
wrote:
> Hi Greg,
>
> TheiProphetpowerful tool that is very simple to run.  Unless you are
> testing the performance of the various modelsiProphetapplies, there
> is not need to setiProphetparameters.  The easiest way to run the

> tool is on a single analysis (one experiment, one search engine),

> this only requires enabling the xinteract -ip option which will runiProphetand ProteinProphet inIPROPHETmode in parallel to the

> standard analysis.
>
> A more complex analysis can be setup for combining multiple
> experiments and multiple search engines. The individual search engines
> and experiments must first be analyzed using xinteract (using option
> -E<experiment_label>)  where the different searches of the same data

> get labeled with the same <experiment_label>.  You can then runiPropheton the PeptideProphet results using the InterProphetParser

> tool.
>
> Running it on the commandline without any arguments produces following
> the usage statement:
>
> usage: InterProphetParser <file1.pep.xml> <file2.pep.xml>... <outfile>
>
> Where each of the input pep.xml files must come from a PeptideProphet
> analysis, and the spectra that were searched with different search
> engines have the same experiment_label spectrum_name combination.  The

> outfile is a name of the pep.xml file whereiProphetwill write the

> results.  For example:
>
> > InterProphetParser exp1/sequest/interact.pep.xml exp1/xtandem/interact.pep.xml exp1/mascot/interact.pep.xml exp2/sequest/interact.pep.xml exp2/xtandem/interact.pep.xml exp2/mascot/interact.pep.xml interact.iproph.pep.xml
>

> To run ProteinProphet on theiProphetresults you can use the command
>
> > ProteinProphet interact.iproph.pep.xml interact.iproph.prot.xml
>
> -David

>
> On Tue, Jun 30, 2009 at 9:33 AM, Greg Bowersock<bowers...@gmail.com> wrote:
>
> > Just bumping this one up, since it looks like it got missed.
>
> > On Jun 10, 12:08 pm, "bowers...@gmail.com" <bowers...@gmail.com>
> > wrote:
> >> I was wondering if someone could post some more detail oniProphet.
> >> One thing that I would like to know is what all command line arguments
> >> it takes, maybe some sample command lines, as that is how I work with
> >> the TPP tools. I've gotten a little behind on some of the newer tools,

> >> but I thinkiProphetis something I've been looking for.
>
> >> Thanks,
> >> Greg

[David Shteynberg]

Hi Bernt,

This was only an example. In my work I organise my searches in
different directories. This is not the only way this tool can be run.
What you suggest is an acceptable alternative.

-David