psidev-pi-dev

[Psidev-pi-dev] Issue 45 in psi-pi: Number of peptides / proteins to be reported

[<cod...@go...>]

Status: Accepted
Owner: dcreasy
Labels: Type-Defect Priority-Medium

New issue 45 by dcreasy: Number of peptides / proteins to be reported
http://code.google.com/p/psi-pi/issues/detail?id=45

Writing documentation always brings a few problems to the surface...

Consider 2 separate use cases:
  - Import into a repository such as Pride
  - Import into an application such as Scaffold or Peptide Prophet

With the first, you probably only want to report and store proteins above a
certain threshold that for example gives an FDR described in the instance
document.
With the second, we need to give scores for all spectra. (See:
http://code.google.com/p/psi-pi/wiki/NotesForDocumentation#How_many_should__be_saved
for details).

One option is to insist that these two cases require two different instance
documents.

Another option is to 'encourage' people to save a larger number of results
and add (say) a boolean attribute to <SpectrumIdentificationItem> and
<ProteinDetectionHypothesis> that indicates that the result is above the
specified thresholds.
Without the boolean flag, if all results are saved (as in the current
Mascot examples), a repository such as Pride would need to understand the
relevant CV and only store results above a given score.

Any thoughts / preferences?



-- 
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings

[Psidev-pi-dev] Issue 45 in psi-pi: Number of peptides / proteins to be reported

[<cod...@go...>]

Comment #1 on issue 45 by andrewrobertjones: Number of peptides / proteins  
to be reported
http://code.google.com/p/psi-pi/issues/detail?id=45

I think we can support this for SpectrumIdentifications with no schema  
change. We
can encourage data producers to produce large number of spectrum  
identifications.
The identifications deemed "correct" are the ones referenced from
ProteinDetectionHypotheses. This then also supports the use case of peptide
identification scores being boosted once a protein detection has been made.  
PRIDE
can make a call on how many of the SpectrumIdentifications to load.

This does not solve the FDR problem for proteins though. However, I think  
the
solution of creating two different instance documents would be okay. If  
sending to
PRIDE, just send the list of proteins found. If using analysisXML as an  
intermediate
format, send a long list of proteins (and peptides) with no set criteria  
for what
is "correct".

-- 
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings

[Psidev-pi-dev] Issue 45 in psi-pi: Number of peptides / proteins to be reported

[<cod...@go...>]

Updates:
	Status: Fixed

Comment #2 on issue 45 by dcreasy: Number of peptides / proteins to be  
reported
http://code.google.com/p/psi-pi/issues/detail?id=45

Added PassThreshold attribute to SpectrumIdentificationItem and
ProteinDetectionHypothesis

--
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings

[Psidev-pi-dev] Issue 45 in psi-pi: Number of peptides / proteins to be reported

[<cod...@go...>]

Updates:
	Status: Accepted

Comment #3 on issue 45 by dcreasy: Number of peptides / proteins to be  
reported
http://code.google.com/p/psi-pi/issues/detail?id=45

I've re-opened this issue rather than add the problem to the end of the CV  
issue
because it seems more relevant here. We (or maybe just me) may have lost  
the plot here...

Comment 2 says what we added to the schema. The structure for specifying  
which items
"PassThreshold" is described here:
http://code.google.com/p/psi-pi/issues/detail?id=49#c3

Currently, the only CV items allowed in the 'Threshold' are:
MS:1001448: pep:FDR threshold
MS:1001447: prot:FDR threshold
MS:1001494: no threshold

However, using an FDR is not the only way to do this. For example, if  
you've only got
a few spectra, then FDR is definitely is not an option. I think we need  
other terms
to be allowed here. For example, you might want to specify that the  
threshold is
"ZYXScore > 23.4"

In the Mascot example, there is for example:
<ProteinDetectionProtocol id="PDP_MascotParser...
   <AnalysisParams>
     <cvParam accession="MS:1001316" name="mascot:SigThreshold"  
cvRef="PSI-MS"
value="0.05"/>

And I guess in this case I would like to specify MS:1001316 in
AnalysisProtocolCollection/SpectrumIdentificationProtocol/Threshold/



--
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings

[Psidev-pi-dev] Issue 45 in psi-pi: Number of peptides / proteins to be reported

[<cod...@go...>]

Comment #4 on issue 45 by andrewrobertjones: Number of peptides / proteins  
to be reported
http://code.google.com/p/psi-pi/issues/detail?id=45

Agreed, we need a few different terms here, not sure if they exist in the  
CV or just
not in the mapping at present e.g. p-value, mascot:SigThreshold, some terms  
for
Sequest etc.



--
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings

[Psidev-pi-dev] Issue 45 in psi-pi: Number of peptides / proteins to be reported

[<cod...@go...>]

Comment #5 on issue 45 by pierrealainbinz: Number of peptides / proteins to  
be reported
http://code.google.com/p/psi-pi/issues/detail?id=45

for Phenyx we probably already have something:

a) representation of peptide scores:

id: MS:1001395
name: Phenyx:Pepzscore
and
id: MS:1001396
name: Phenyx:PepPvalue
both are
is_a: MS:1001143 ! search engine specific score for peptides
is_a: MS:1001153 ! search engine specific score


id: MS:1001384
name: Phenyx:MinPepzscore
and
id: MS:1001385
name: Phenyx:MaxPepPvalue
both is_a: MS:1001302 ! search engine specific input parameter


2) and we also have 2 binary values for a "valid" or "accepted" peptide  
status,
corresponding to automatic selection (unedited search result) and  
user-defined
selection (that has gone through manual selection), respectively:
id: MS:1001393
name: Phenyx:Auto
and
id: MS:1001394
name: Phenyx:User
both defined
is_a: MS:1001143 ! search engine specific score for peptides
is_a: MS:1001153 ! search engine specific score


As the thresholding we talk about about is a "post processing" event  
(something we
apply to the "raw" result), which one would you consider? I'm just  
wondering which
one could fit best for a passThreshold criterium.


--
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings

[Psidev-pi-dev] Issue 45 in psi-pi: Number of peptides / proteins to be reported

[<cod...@go...>]

Updates:
	Status: Fixed
	Labels: Milestone-Release1.0

Comment #6 on issue 45 by eisenachM: Number of peptides / proteins to be  
reported
http://code.google.com/p/psi-pi/issues/detail?id=45

(No comment was entered for this change.)

--
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings