Mass deviation correction tool?

[Steven Verbruggen]

Hi all,
I am currently searching public data using MS-GF+ and X!Tandem, but I had unexpectedly very few identifications. Unfortunately, it appeared that the public data was generated on a mass spectrometer with quite some mass bias (see figure in attachment). Originally, I ran the search with 6 ppm precursor tolerance. As a proof-of-concept, I ran the search also with 15 ppm precursor tolerance and this made the identification rate to raise drastically. Of course, it would be much better to ‘recalibrate’ the data (I suppose something like a linear shift) in order to shift the average mass error to 0, so that I can do the search with a normal 6ppm precursor tolerance. I know that tools like PEAKS and MaxQuant do this automatically, but it is important for the downstream analysis to use MS-GF+ and X!Tandem. Anyone who knows a tool which can perform this task prior to the MS-GF+ and X!Tandem searches?
Thanks for your suggestions!

[sachs...@gmail.com]

Hi,

you could use the OpenMS tool https://abibuilder.informatik.uni-tuebingen.de/archive/openms/Documentation/release/latest/html/TOPP_InternalCalibration.html and feed in high confidence ids from some prior search (option -cal:id_in ).

Make sure that you have a target/decoy database (e.g., create with DecoyDatabase -in ${FASTA} -out ${FASTA_TD} )

Then you could get the high-confidence ids with:

  MSGFPlusAdapter -in ${f} -out ${f}.idXML -database ${FASTA_TD} 

  PeptideIndexer -in ${f}.idXML -out ${f}.idXML -fasta ${FASTA_TD} -enzyme:specificity none

  FalseDiscoveryRate -in ${f}.idXML -out ${f}_001.idXML -FDR:PSM 0.01

and feed them into the InternalCalibration tool

[mva...@gmail.com]

Hi,

I would also recommend using OpenMS for this. If you want a "Plan B", you can export recalibrated spectra from PeptideShaker. Via the "Export -> Follow-up Analysis" menu from the GUI or using the FollowUpCLI (https://github.com/compomics/peptide-shaker/wiki/PeptideShakerCLI#c---followupcli). Note that currently only mgf is supported as output format and we did not evaluate the performance of the recalibration in a long time. If you give it a go I would be curious to know how well it performs in your hands.

Hope it helps,

Marc

On Tuesday, June 15, 2021 at 9:47:41 AM UTC+2 stevenver...@gmail.com wrote:

[RayS]

MS-Fragger  too can export calibrated as well as uncalibrated MGF which you can use further. I guess ProteomeDiscoverer too can do that.