I know there's been research on predicting protein structure from peptide sequences AFTER they've been identified from MS/MS spectra, but I am trying to track down any papers that have tried to infer protein structure directly from the spectra signal itself? I couldn't find much from searching online, but I have a CS background so I may just not be searching with the right terminology. After reading papers that compare different peptide search engines and seeing how vastly different identification results can be using the same data, I'm wondering if it would be better/possible to infer a protein's secondary structure from the MS spectra directly.
I have done some proteomic work in the past so I will report some few thinks I learned.
I do not really know what kind of MS data you have, but in my case I hade LC-MS/MS data. The protein were separeted via SDS, the gel was then sliced and protein were digested. The peptide mixture was then ran via LC and finally an MS/MS was performed on them. What you get from the MS/MS is usually a fingerprint that allows you to reconstruct the aminoacidic sequence of your peptides. After this you will have to use programs, such as MASCOT or TPP (http://www.matrixscience.com/, http://www.proteomecenter.org/software.php), for assigning these peptides to proteins. Now, in case you work with an organism that as a genome available the assignment of the protein can be relativelly feasible. More tricky I think it will be to assign the peptides to proteins of unknown origin. However, there are pleanty of paper out there that achieved that.
To infer the structuere (II or III) there are some specific MS techniques you maight want to perform (http://en.wikipedia.org/wiki/Protein_mass_spectrometry) or X-ray crystallography. In case you have a workflow similar to the one I described above, meaning you get from the MS/MS to the aa sequence, you might want to consider to use a tool that predicts the secondary structure starting from aa seq. You can find a short description here: Protein secondary-structure prediction
Hope this helps.