MyriMatch error (Version: SearchGUI 3.3.17)

[Juan Camilo Rojas Echeverri]

Dear support,

I tried running a search using MyriMatch and I keep getting this error:

Mon Feb 03 18:38:17 CET 2020        Indexing spectrum files.
Mon Feb 03 18:38:17 CET 2020        Extracting search settings.

Mon Feb 03 18:38:17 CET 2020        Error: 20
Mon Feb 03 18:38:17 CET 2020        An error occurred while running SearchGUI. Please contact the developers.
Mon Feb 03 18:38:17 CET 2020        The search or processing did not finish properly!

Mon Feb 03 18:38:17 CET 2020        Search Canceled!

And this is written in the log:

java.lang.ArrayIndexOutOfBoundsException: 20
    at eu.isas.searchgui.processbuilders.MyriMatchProcessBuilder.getVariableModificationsAsString(MyriMatchProcessBuilder.java:406)
    at eu.isas.searchgui.processbuilders.MyriMatchProcessBuilder.<init>(MyriMatchProcessBuilder.java:159)
    at eu.isas.searchgui.SearchHandler$SearchWorker.doInBackground(SearchHandler.java:2027)
    at javax.swing.SwingWorker$1.call(Unknown Source)
    at java.util.concurrent.FutureTask.run(Unknown Source)
    at javax.swing.SwingWorker.run(Unknown Source)
    at java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source)
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
    at java.lang.Thread.run(Unknown Source)

Any suggestions?

Sincerely,

Juan C.

[Harald Barsnes]

Hi Juan C.,

It seems like you are using more than 20 variable modifications?! Is that correct? And if so, why? :)

Generally, searching with this many variable modifications is not a good idea, given that the matches between the spectra and the peptide sequences will then be more or less random, as you give the search engines way too much freedom in the matching process (or perhaps more specifically: room to make mistakes).

Best regards,

Harald

[Juan Camilo Rojas Echeverri]

Hi Harald,

I work specifically with Carbonylation PTMs which requires to make searches for many modifications at the time.

With regards what you mention about the room to make mistakes, I have observed this to be true for PSMs to very noisy spectra and big proposed "peptides" (sometimes just random m/z noise values selected for "fragmentation". Which is why I wanted to include Myrimatch since it includes an itensity score in its algorithm (at least this is what I understood from a quick check on Chambers paper).

Nevertheless, I have been using other search engines (X!Tandem, MS-GF+, MSAmanda) through SearchGUI (aware of going against the suggestions) using the same number of modifications. Through spectral validation, specific reporter ions, modification specific neutral losses, and cross-correlation of results obtained with PEAKS I am confident that it has performed well for good MSMS. However, I can't comment on the statistical validation results. I only use spectra I have checked for my results.

I think this might be a debatable idea and don't have a strong lenience to any side, but premises of search engines like ionbot, PEAKs or other OMS argue that considering more PTMs or, in the lack of its identity, mass shifts at the same time would give a more confident match. Reducing the number of false negatives and false positives (although as you mention this can be the opposite for badly curated MSMS spectra). Any further comments on this or suggested reading?

With regards to Myrimatch then I should just understand that is a search engine limitation on the number of PTMs considered per search?

Sincerely,

JC

--
You received this message because you are subscribed to the Google Groups "PeptideShaker" group.
To unsubscribe from this group and stop receiving emails from it, send an email to peptide-shake...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/peptide-shaker/f6c822de-ccd5-4a8e-9aeb-1e4cb0521830%40googlegroups.com.

[Harald Barsnes]

Hi JC,

I'm not too familiar with carbonylation PTMs, so maybe I can ask why these would require searching for more than 20 different variable PTMs? Is it not possible to merge some of these? Perhaps can share your search parameter file so that I can get a better understanding?

I will leave it to others to comment on the validity of the search results when this many variable PTMs are used. Hopefully other users have more experience with this approach?

Regarding the specific MyriMatch issue, it's due MyriMatch representing each variable modification with a unique single character, which cannot be any of the letters representing amino acids. At the moment this list (as provided by SearchGUI) consists of 20 options, hence it breaks down when you use more than 20 different variable modifications at the same time.

Best regards,

Harald

[Juan Camilo Rojas Echeverri]

Hi Harald,

Here I attach the parameters file I have been using. The PTMs were based on derivatization products with a carbonyl-specific biotin tag (ARP) of metal catalyzed oxidation carbonyl PTMs reported previously (10.1016/j.jprot.2016.12.019) and a few other carbonylation PTMs we look for in our lab. This list is very limited with respect to all the possible products that could be generated in-vivo, but I had to compromise with those for now.

Thanks for the explanation with respect to Myrimatch.

Sincerely,

JC

--

You received this message because you are subscribed to the Google Groups "PeptideShaker" group.
To unsubscribe from this group and stop receiving emails from it, send an email to peptide-shake...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/peptide-shaker/c3d94fbe-2c51-4d5b-b053-382091ac8a88%40googlegroups.com.