Need help for targeted MS/MS aquisition

[karthikskamath]

Hello,

I have run sucrose gradient fractionated (acetone precipitated) (HEK) cell lysed proteins on (6x8 cm dimension) 1D-SDS-PAGE. The gel was sliced and in-gel digested (with alkylation and reduction) . Digested peptides were analyzed Agilent (1200 series) Nano-LC and LTQ-Orbitrap (Discovery) systems. We have performed 110 minutes RP-gradient to separate peptides. A top 10 peaks selection based DDA (with CID) was used for acquisition.

When we analyzed the results using MASCOT and Swissprot, as anticipated we generated a list of proteins. But we were unable to find back about 12 signature proteins associated with specific biology question. These proteins are used as markers while performing westerns and antibody labeled imaging. The proteins might have gone undetected due to its low abundance and problem with dynamic range of the instrument.

Hence I planned to perform more targeted acquisition using a inclusion list (to be active while choosing the masses for MS/MS acquisition). I took the help of the tool MS-Digest . I performed the in-silico digest of candidate proteins. After picking +2 and +3 monoisotopic masses to be used, each protein is generating around 200 MI masses, but limit of number of parent ion inclusion list is about 500 masses per method. Hence it is problem to analyze all the candidate proteins in a single run.

My question is; I wish to scrutinize the list of peptides/masses to be used in parent ion inclusion list. I can think of some of criterion of peptides for selection such as hydrophobicity, property of fragmentation etc. I am seeking for a tool which does this selection in-silico and provides a scoring kind of gauge as a measure of probability of the peptide being analyzed. Can any one in the forum suggest with such tool published?

[Doug]

I suspect that there is some MRM software that could probably help you out here. I don't have any experience with this. But you could look into Skyline for starters.

For a very low throughput solution you could look at Peptide Atlas. You can look up your proteins and see which peptides have been observed in the past and how often. Good luck. And please tell us if you find a good solution.

[Craig]

For a more automated solution you could try PeptideSieve. You can run all your proteins through that and then select the top n most proteotypic peptides from each. For 12 proteins I would use an n of 20. Then calculate the expected m/z for charge states 2 and 3 (maybe 4 with a missed cleavage), and load those onto your inclusion list.

[karthikskamath]

Thanks Craig and Doug. I am going through related papers. Sure I will share the final experimental set-up and outcome. I am keeping my fingers crossed

[karthikskamath]

Hello Craig,

I have run the sequence through PeptideSeive.

I am getting List of peptides.jpg. What I am not clear about is what does the fields marked in red in this image mean? What is the scale of score given? Is the score near to one good or it is other way round? I wish to further use the score for sorting the peptides.

Thanks in advance...

[Craig]

I am not sure what the red means because I have always only used the command-line version available here. Are you using a web version? I have also only provided pre-digested peptides as input with the TXT format, never full proteins in a FASTA. For some reason it looks like the peptides it is generating are all shifted by one residue. In terms of score, it is normalized between 0 and 1 where higher means more likely to be observed. I use the MUDPIT_ESI properties and in my experience, with unfractionated E. coli, anything with a score above 0.75 is significantly more likely to be observed.

[karthikskamath]

Hello Craig,

Many thanks. I have used user interface based version available here. I will consider the score threshold. Since the list I have generated is for about 15 proteins is less than 300 (upper limit of inclusion list masses is ~500) I have been inclusive in adding some more which are below 0.75 threshold. Thanks again.

[proteovations]

I would use skyline (and even think about spending enough time to be proficient with it) because it is becoming very popular and is great software. No I do not work for Mike or the developers of skyline

If you still have proteins you do not see, I would set up a MRM style experiment with the LTQ-Orbitrap (just because I know how to use it and I know it scans fast). Use several peptides for each protein you don't see a CID scan mass and set the +2 mass for MSMS data (unless it is a big peptide and then use +3 instead). I can explain better if you are interested.

The way you are performing the experiment now will allow you to potential ID the proteins but the MRM style experiment will let you ID them for sure if they being ionized at all. If you have 12 proteins and monitor the top 2 peptides for each, than that would be 24 MSMS scans in a row. On an LTQ XL, I would guess that would take about 3-4 seconds for each cycle which is plenty of time to detect something even if your peak width is only 10 seconds (most people peaks are wider than this). Once you have IDed some of them in a run, you can go back and add more peptides for other proteins you did not ID yet.

Skyline can help you set up the MSMS instrument file.

Just my 2 cents.