SearchGUI and PeptideShaker for large database search.

[Pratik Jagtap]

Hello Marc and Harald,

I am trying to set up searches using SearchGUI (MS-GF+) and PeptideShaker (outside of GalaxyP) using a target database only. Is there a way that I can do this using SearchGUI and generate a spectrum or peptide report using PeptideShaker?

I attach two messages that showed while trying to set up a target search.

Your inputs will be appreciated.

Thanks and Regards,

Pratik

Pratik Jagtap,

Managing Director,

Center for Mass Spectrometry and Proteomics,

43 Gortner Laboratory
1479 Gortner Avenue
St. Paul, MN 55108

Phone: 612-624-9275

[Marc Vaudel]

Hi Pratik,

You should be able to run a search with target only after ignoring all the warnings, yes. Note that we did not test this use case in a long time, but if you encounter any problem we will be happy to help :)

Using target only however kind of ruins the point of using PeptideShaker who strongly relies on the decoy to evaluate the quality of matches and build peptides and proteins. As said in the warnings, without decoy, we cannot merge results from different search engines, the protein inference will not be as accurate and there will be no quality metric for the matches. Can you give me more input on why you need to do so?

Best regards,

Marc

[Pratik Jagtap]

Hello Marc,

> Can you give me more input on why you need to do so?

Thanks - I will test this. For our metaproteomic and proteogenomics searches we have been using the two-step method ( ). In this method, we search with target database first, parse out sequences that have been identified in the first search and use it for a target-decoy database search in second step. This has resulted in an improved sensitivity of identifications because of reduced database size. In our final step and workflow, we use PSM evaluation to test and visualize spectral annotations to ensure that our identifications can be reported as novel proteoforms (or metaproteins). 

In our current workflows,  we use ProteinPilot (a commercial search algorithm) as a search algorithm would like to replace it with SearchGUI-PeptideShaker. 

Regards,

Pratik

Pratik Jagtap,

Managing Director,

Center for Mass Spectrometry and Proteomics,

43 Gortner Laboratory
1479 Gortner Avenue
St. Paul, MN 55108

Phone: 612-624-9275

[Pratik Jagtap]

Here is the reference for two-step method - http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3633484/

We have also used this for our recent proteogenomics manuscript - z.umn.edu/pgfirstlook

- Pratik

Pratik Jagtap,

Managing Director,

Center for Mass Spectrometry and Proteomics,

43 Gortner Laboratory
1479 Gortner Avenue
St. Paul, MN 55108

Phone: 612-624-9275

[Marc Vaudel]

Hi Pratik,

Thank you for the information, that looks very promising. If you are using PSM level results with a single search engine, using target only should not impair your results as far as I can see. If you encounter any issue, don't hesitate to contact us and we will fix it. Also, if you encounter performance problems, in terms of speed or memory, I can look into it - the tool was not really designed for such gigantic databases :-)

PeptideShaker has various metrics for the evaluation of the quality of matches which we use on top of the target/decoy approach (the difference between green and yellow hits). These include precursor mass deviation, coverage of the sequence by fragment ions, spectrum annotation, etc. I guess these would be relevant in the first step of your process. You will find them when designing your export after clicking on "see advanced features". If you are missing anything there which could be helpful to evaluate the quality of a PSM, we will be happy to implement it.

Best regards,

Marc