X!tandem error

[zeyu sun]

Hi, Dear SPC users and developers:
I'm wondering if anyone have some clue or remedy to the problem I have with the X!tandem search in TPP(v4.7.1):
I have some Orbitrap Velos data in RAW and converted them to mzXML files by MSconverter (ProteoWizard).
I ran the Comet pipeline on them and everything was just fine.
However, when I search the same mzXMLs by Tandem, some of those files gave error reports:
_________________________________________
X! TANDEM Jackhammer TPP (2013.06.15.1 - LabKey, Insilicos, ISB)

Loading spectra (mzParser)............................ loaded.
Spectra matching criteria = 55267
Pluggable scoring enabled.
Starting threads . started.
Computing models:
    12345678
*********
Warning: One or more sequences appear to contain invalid amino acid codes.

*********

9    sequences modelled = 9 ks
Model refinement:
    partial cleavage .......... done.
    finishing refinement ... done.
Creating report:
    initial calculations  ..... done.
    sorting  ..... done.
    finding repeats ..... done.
    evaluating results ..... done.
    calculating expectations ..... done.
    writing results ..... done.

command "c:\Inetpub\tpp-bin\tandem c:/Inetpub/wwwroot/ISB/data/xxx.tandem.params" failed: Unknown error

Command FAILED
RETURN CODE:65280
 _________________________________________
Despite this error message, the .tandem files were generated,
When I use the command tandem2XML to convert the .tandem files to .pep.xml, as expected, it didn't work out:
Operation not permitted
Command FAILED
 RETURN CODE:256

It only happens in few mzXML files, not all of them. And Comet runs OK on all files, so I guess the mzXMLs should be fine.
I further checked .tandem files, looks like they contain all necessary parts of Xtandem search results, and all spectra-peptide matches were listed there with scores. I just don't know why such files still gave error message and cannot go through tandem2XML.
Please help!
Thank you very much!

SunSun

[Jimmy Eng]

I don't know what the issue is with X!Tandem but just wanted to comment on an unrelated issue which is that you should think twice about using X!Tandem's refinement mode, especially if you are going to push those search results through PeptideProphet or even simple target-decoy FDR.  All proteins that make it to the refinement step (and their corresponding peptides identified before or after refinement) will all have distorted/high statistics.  Here are two publications that discuss this problem:

http://www.ncbi.nlm.nih.gov/pubmed/19947654

http://www.ncbi.nlm.nih.gov/pubmed/21288048

The simple solution is to not use refinement mode.

--
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 http://groups.google.com/group/spctools-discuss.
For more options, visit https://groups.google.com/d/optout.

[zeyu sun]

thank you Jimmy.
I will try the search w/o refinement.
I will let you know if the problem solved.

SunSun

[Eric Deutsch]

I think the key part in your output is this:

 

*********
Warning: One or more sequences appear to contain invalid amino acid codes.

*********

I think this is a warning that we put in to warn that X!Tandem may crash because of your database. You should check your database for validity. Perhaps there are some strange characters in one of the sequences that causes X!Tandem to crash sometimes, but not Comet. We have seen this before.

 

Eric

[zeyu sun]

Thank you Eric,
Is there any tool out there to check the validity of my fasta?

To Jimmy:
I just tried the tandem search without refinement, the problem is still there.

Thank you all for the reply!

SunSun

[Brian Hampton]

Hi SunSun,

I ran a search today and got the same error.  I was using a bovine database that I downloaded quite a while ago.  So I downloaded the latest version today, appended the CRAP database to it and then ran it through the decoy utility to create a concatenated forward and reversed sequence database.  Subsequent searches with the new database completed without error.

I can only presume the older database became corrupted somehow.  I don't know of a utility that will validate the integrity of a database.  If you know regex or maybe perl you could write a script to look for characters that are not part of the amino acid code.  For me the expedient solution was to download a fresh version of the database.

Brian

[zeyu sun]

OK, the problem solved! I updated to the latest uniprot fasta and the tandem and tandem2XML work fine now.
thankyou all!
SunSun