top two peptides with the same score in the same scan, which one to choose?

Search algorithms, post-searching processing, quantitation software, etc. Share and discuss software here.
isobaric
Angiotensin Member
Angiotensin Member
Posts: 42
Joined: Fri Aug 10, 2012 9:06 pm

top two peptides with the same score in the same scan, which one to choose?

Postby isobaric » Sat Oct 05, 2013 11:31 am

Hi everyone,

Database searching engine usually would assign a score to each of these candidate peptides and choose the one with the highest score as the identified peptide for a MS/MS spectrum. If there are two candidate peptides with the same highest score, is there any rule regarding which one to choose as the identified peptide? Or just pick up one randomly?

Thanks

Craig
E. Coli Lysate Member
E. Coli Lysate Member
Posts: 220
Joined: Sun Jun 26, 2011 6:49 pm

Postby Craig » Mon Oct 07, 2013 6:58 am

If you're trying to decide between a target and a decoy peptide, there is a thread (and poll) which addresses that issue here. If you're trying to decide between two target peptides I don't think there is any rule to help you. It's possible one of the recent protein grouping algorithms addresses this case. I think picking one randomly is your best bet.

User avatar
Doug
E. Coli Lysate Member
E. Coli Lysate Member
Posts: 307
Joined: Sun Jun 26, 2011 7:20 pm

Postby Doug » Mon Oct 07, 2013 9:17 am

I don't think it matters. If that happens it means you haven't really identified the peptide at all. But if your FDR filtering is working properly than this (in combination with other sorts of false positives) should occur at a rate below your FDR. It seems like something that should occur pretty infrequently and should be accounted for with target/decoy searching. I wouldn't worry about it too much.
"If we knew what we were doing it wouldn't be research." -AE

isobaric
Angiotensin Member
Angiotensin Member
Posts: 42
Joined: Fri Aug 10, 2012 9:06 pm

Postby isobaric » Mon Oct 07, 2013 7:28 pm

Craig and Doug, Thanks for your reply.
I was referring to two target peptides with identical score. I think this would actually happen very frequently in the PTM search. For example, there is a mono-phosphorylated peptide XXXSSSXXX that has three serine next to each other. I would say its really hard to determine whether the phosphorylated peptide is in the form of XXXpSSSXXX, XXXSpSSXXX, or XXXSSpSXXX because these three Serine residues are so close to each other and the chance of seeing these site-determining fragment ions to determine the correct modification site is really small. If you fails to see these site-determining fragment ions, these three possible phosphopeptides should recieve the same score(assuming this score is the highest score). If the database searching software picks peptide randomly and if this peptide has been identified in many spectra, it is highly likely that all three phosphopeptides will be reported in the final identification result. However, there is only one in reality. A good PTM site localization algorithm should be able to at least help make a conservative call (one "unlocalized" phosphorylated site, instead of three phosphorylated sites), but this is certainly not the case as shown in the Marx's et al A large synthetic peptide and phosphopeptide reference library for mass spectrometry-based proteomics. All current popular site localization software tested in that paper underestimated false localization rate.
But, a phosphopeptide with three Ser next to each other may not happen in real biological system

Craig
E. Coli Lysate Member
E. Coli Lysate Member
Posts: 220
Joined: Sun Jun 26, 2011 6:49 pm

Postby Craig » Tue Oct 08, 2013 6:47 pm

PTM searches are a different story. Without localization you definitely risk over-reporting PTM sites, as you mentioned. When you're not searching PTMs that frequently occur multiple times per peptide I don't think it's much of an issue.


Return to “Bioinformatics”

Who is online

Users browsing this forum: Google [Bot] and 1 guest