Problem of model fitting (PeptideProphet) using MS-GF+ results

[Honglan Li]

Hi, everyone.

I want to run PeptideProphet using MS-GF+ results.

But it's not works well. 

In order to run PeptideProphet, we converted *.mzid to pepXML using idconvert (tpp 4.8 , ubuntu 14.04)

Each psm have five types of scores:

-raw

-denovo

-SepcEValue

-IsotopeError

Then we executed the PeptideProphet with coverted file (cmd: xinteract -dXXX_ *.xml) and received such errors as follows:

Would you please give me some advises for solving this problem?

Thanks a lot.

----

running: "/usr/local/tpp/bin/PeptideProphetParser 'interact.pep.xml' DECOY=XXX_"

Using Decoy Label "XXX_".

 (MS-GF+)

WARNING!! The discriminant function for MSGFPL is not yet complete.  It is presented here to help facilitate trial and discussion.  Reliance on this code for publishable scientific results is not recommended.

WARNING: MSGFPL only support semi-parametric PeptideProphet modelling, which relies on a DECOY search.

WARNING: MSGFPL only support semi-parametric PeptideProphet modelling, which relies on a DECOY search.

WARNING: MSGFPL only support semi-parametric PeptideProphet modelling, which relies on a DECOY search.

WARNING: MSGFPL only support semi-parametric PeptideProphet modelling, which relies on a DECOY search.

WARNING: MSGFPL only support semi-parametric PeptideProphet modelling, which relies on a DECOY search.

WARNING: MSGFPL only support semi-parametric PeptideProphet modelling, which relies on a DECOY search.

WARNING: MSGFPL only support semi-parametric PeptideProphet modelling, which relies on a DECOY search.

init with MS-GF+ trypsin 

Segmentation fault (core dumped)

 

[Honglan Li]

My coverted file is here: https://drive.google.com/file/d/0B5i2NDdAgrJ3Y2diWDAzVHBQSm8/view?usp=sharing

2016년 3월 24일 목요일 오후 5시 7분 32초 UTC+9, Honglan Li 님의 말:

2016년 3월 24일 목요일 오후 5시 7분 32초 UTC+9, Honglan Li 님의 말:

[David Shteynberg]

MSGF+ data can be processed using the semi-parametric model as written in the WARNING messages.  To enable you must use -OP when running xinteract.  I suggest you try the following xinteract command:

xinteract -OPd -p0 -dXXX_ *.xml

--
You received this message because you are subscribed to the Google Groups "spctools-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to spctools-discu...@googlegroups.com.
To post to this group, send email to spctools...@googlegroups.com.
Visit this group at https://groups.google.com/group/spctools-discuss.
For more options, visit https://groups.google.com/d/optout.

[Honglan Li]

Thanks for your quick reply. 

   I tried the command you told me, but it stll didn't work well.

   

   The warning msg is "The discriminant function for MSGFPL is not yet complete" 

   How to fix it ?

   

  cmd: 

  xinteract -OPd -p0 -dXXX_ *.xml

 Error msg I received

  ----

running: "/usr/local/tpp/bin/DatabaseParser 'interact.pep.xml'"

command completed in 0 sec 

running: "/usr/local/tpp/bin/RefreshParser 'interact.pep.xml' 'C:\Users\JONGHUN\Desktop\MSGFPlus\C-25fmol\C-25fmol-R1\1. CYH\uniprot_C+H+Y+48.revCat.fasta'"

  - Building Commentz-Walter keyword tree...  - Searching the tree...opening "C:\Users\JONGHUN\Desktop\MSGFPlus\C-25fmol\C-25fmol-R1\1. CYH\uniprot_C+H+Y+48.revCat.fasta" as "/home/mllab/hllee/uniprot_C+H+Y+48.revCat.fasta"

  - Linking duplicate entries...  - Printing results...

command completed in 11 sec 

running: "/usr/local/tpp/bin/PeptideProphetParser 'interact.pep.xml' NONPARAM DECOYPROBS MINPROB=0 DECOY=XXX_"

Using Decoy Label "XXX_".

Decoy Probabilities will be reported.

Using non-parametric distributions

 (MS-GF+) (minprob 0)

WARNING!! The discriminant function for MSGFPL is not yet complete.  It is presented here to help facilitate trial and discussion.  Reliance on this code for publishable scientific results is not recommended.

init with MS-GF+ trypsin 

Segmentation fault (core dumped)

command "/usr/local/tpp/bin/PeptideProphetParser 'interact.pep.xml' NONPARAM DECOYPROBS MINPROB=0 DECOY=XXX_" exited with non-zero exit code: 35584

QUIT - the job is incomplete

-----

XML format per PSM: (file: https://drive.google.com/file/d/0B5i2NDdAgrJ3Y2diWDAzVHBQSm8/view?usp=sharing)

  <spectrum_query spectrum="C:\Users\JONGHUN\Desktop\MSGFPlus\C-25fmol\C-25fmol-R1\1. CYH\C-25fmol-R1_QEx2_000090.51174.51174.4" spectrumNativeID="index=51174" start_scan="51174" end_scan="51174" precursor_neutral_mass="3972.990669523105" assumed_charge="4" index="1">

      <search_result>

        <search_hit hit_rank="1" peptide="VGNVGEDDAIPEVSHAGDVSTTLQVVNELLKDETVAPR" peptide_prev_aa="R" peptide_next_aa="F" protein="sp|P16521|EF3A_YEAST" num_tot_proteins="1" calc_neutral_pep_mass="3972.99140194498" massdiff="7.32421875e-04" protein_descr="sp|P16521|EF3A_YEAST Elongation factor 3A OS=Saccharomyces cerevisiae (strain ATCC 204508 / S288c) GN=YEF3 PE=1 SV=4" num_tol_term="2" num_missed_cleavages="1">

          <search_score name="raw" value="269"/>

          <search_score name="denovo" value="289"/>

          <search_score name="SpecEValue" value="2.8337685502588284E-39"/>

          <search_score name="EValue" value="1.2856572993111488E-31"/>

          <search_score name="IsotopeError" value="0"/>

        </search_hit>

      </search_result>

    </spectrum_query>

[David Shteynberg]

Hello Honglan,

I was unable to replicate the error on my system, however, I have tested that this command works on a more recently compiled version of PeptideProphet.  You can download the binary here https://dl.dropboxusercontent.com/u/21286225/PeptideProphetParser.exe and run as follows

PeptideProphetParser.exe test.pep.xml NONPARAM DECOY=XXX DECOYPROBS MINPROB=0 CLEVEL=2 ACCMASS PPM

I had to set a fairly high CLEVEL value on this data since MSGF+ suffers from scoring highly of some decoys and is not always accurate at the critical 1% error.  The WARNING message in the new code has been updated to reflect this problem.

Cheers,

-David

[Honglan Li]

Thank you very much for your help, David.

It works well now.

Best wishes,

Honglan Li

2016년 3월 26일 토요일 오전 2시 33분 59초 UTC+9, David Shteynberg 님의 말:

[Honglan Li]

Hi David.

It's me again.

Last time, you gave me the s/w and cmd. 

It's works well when I use the sample I gave you, and some other XML files.

But for some XML files, it still do not work well.

                  XML file: https://drive.google.com/file/d/0B5i2NDdAgrJ3bDY4ZFR0M2VOUlU/view?usp=sharing

When I executed this test sample (test2.pep.xml.xml) with the cmd you told me

                "PeptideProphetParser.exe test2.pep.xml.xml NONPARAM DECOY=XXX DECOYPROBS MINPROB=0 CLEVEL=2 ACCMASS PPM"

I received msg as follows: (detail: I added captured error msg file)

----

KDModel.cpp, Line 144

Expression: idx < posprobs_ -> size()

This application has requested the Runtime to terminate it in an unusual way.

Please contact the application's support team for more information

----

Would you please tell me what happened to this XML file?

Thanks a lot.

Honglan Li

2016년 3월 28일 월요일 오전 8시 52분 13초 UTC+9, Honglan Li 님의 말:

[Honglan Li]

Hi David

It's me again.

Thanks for giving me new PeptideProphet file to run MS_GF+ results.

But it still has problem to execute new test file. 

Would you please tell me the reason?

Thanks a lot.

Honglan Li

XML file: https://drive.google.com/file/d/0B5i2NDdAgrJ3bDY4ZFR0M2VOUlU/view?usp=sharing

Recieved msg:

----

KDModel.cpp, Line 144

Expression: idx < posprobs_ -> size()

This application has requested the Runtime to terminate it in an unusual way.

Please contact the application's support team for more information

----

Thanks a lot.

Honglan Li

2016년 3월 30일 수요일 오후 8시 30분 5초 UTC+9, Honglan Li 님의 말:

[David Shteynberg]

Hi Honglan,

I am on vacation this week. I did fix another processing issue with MSGF+ earlier, but I was still testing.  I will get back to you once I'm back next week.

David

[David Shteynberg]

Hello Honglan,

It appears that the paths to your data contain some restricted characters.  The only allowed characters when naming paths that are compatible with the TPP are alpha-numeric and '_' (underscore) ('.' and whitespace are disallowed.)  That doesn't mean that it won't work sometimes when the paths contain restricted character, but it can be hit or miss.   Please rename the directories and redo analysis.

-David

  

[Honglan Li]

Thanks a lot.

These days, I tested all of my sample data as you told me.

ALL of them worked well.

Thank you for helping me.

---One more thing:

I use idconvert to change mzid file to pepXML fomat.

The scan information in changed pepXML is not equal to real scan information.

I checked the file, and found that the scan information in changed PepXML file is same as the 'index" of MGF file.

Am I right?

I just want to confirm it.

Honglan Li.

2016년 4월 19일 화요일 오전 2시 44분 21초 UTC+9, David Shteynberg 님의 말:

[Honglan Li]

Hi David,

Thank you so much again for sending me the new version of PeptideProphet.

It runs well with MSGF+.

Now, we are comparing other modes of PeptideProphet including unsupervised. Thus, I am just wondering if I can use PeptideProphet in unsupervised mode with MSGF+.

All the best,

Honglan

2016년 5월 2일 월요일 오후 1시 54분 39초 UTC+9, Honglan Li 님의 말:

[David Shteynberg]

Hi Honglan,

Thanks for using the tools!

In my experience, searches against databases that do not contain decoys are fraught with false positives and are not recommended.  Unsupervised support exists for legacy support of certain search engines, however, even in those cases you are less likely to arrive at false results if you include decoys in the database.

Cheers,

-David

[Jonghun Park]

Hi,

I've just found this post while I'm struggling with the same problem.

However, the link above (https://dl.dropboxusercontent.com/u/21286225/PeptideProphetParser.exe) was broken.

Could I get a new link or website to get the file for MSGF+?

Thanks,

Jonghun Park

[David Shteynberg]

Hi Jonghun,

The link to dropbox was never meant to be permanent.  The changes in that binary should have been integrated into the release version of PeptideProphet, installed as part of TPP 5.1.0  Which version of the TPP are you running?

Thanks,

-David

To unsubscribe from this group and stop receiving emails from it, send an email to spctools-discuss+unsubscribe@googlegroups.com.
To post to this group, send email to spctools-discuss@googlegroups.com.

[Jonghun Park]

Oops, I've been using an older version. I will try the latest version.
Thanks for your quick reply!

- Jonghun