Label Free - E. Coli

For this walkthrough, we will continue using data from our paper on Stellar MS with Adaptive RT and PRM Conductor. This section uses a dilution of E. Coli in a HeLa background.

• Open the file LabelFree / ecoli_small_dilution_skyline.sky.
• If you haven't installed the new PRM Conductor v1.1, do that
• Run Expert Review
• If you just did the absolute quantitation walkthrough, then Expert Review has the reference for that experiment, and is set to Isotope Label Heavy. We have to update this.

• Set Isotope Label to Not-Specified
• Press the Current Button. The reference grid is populated now with the replicates from this dilution. You could choose a replicate from this experiment to use as a reference, but I find it more convenient to use a separate reference.
• In the label free scenario, we have the advantage that we likely imported search results into Skyline, so all of the peaks likely are chosen correctly in some initial data set. The only problem is that the quality of these data may be low, due to the low injection times often used for a discovery DIA experiment.
• What I do is use the GPF as a reference to pick the peaks in PRM validation experiments, and then eventually use a PRM experiment as the reference for the remaining experiments in a study.
• Press the Select button, and choose LabelFree / ecolihela_gpf.sky. There are 6 LC injections that comprise this data set, but they were imported as a multi-injection replicate to Skyline, so they show up as a single replicate here. If you had imported the data as 6 single replicates, you could select all 6 of them in the reference grid here.

• Press Start to process the data, and then press the Send button to send the integration boundaries to Skyline.

• Have a look at the worst precursors using the Right key. There are a few that should never have been included in the assay. Let's look at the worst precursor according to the number of outliers, YQGLVAQIELLK. Most of the integration boundaries are over near a small peak at the boundary. If we open the GPF Skyline document and find this peptide, we'll see that it has many more transitions.

• Let's see if we can get better results by using a better reference, from PRM data. Press Select and choose the file ecoli_small_reference.sky, press Start, and then Send when the processing finishes. Use Ctrl+F and paste in YQGLVAQIELLK and search for it. Now these data look very nice, because the reference and dilution experiment data are very similar.

If you look in the Skyline Immediate Window log, you'll see some messages of the type:

2025-05-20 10:22:15.7649|WARN|Thermo.TNG.PeakPick.Algorithm.Logger|No reference found for precursor Sequence: QIIIATGEGAK

If you go to the reference file, for some reason this peptide is not found. It looks like there was an acquisition issue with this one, potentially caused by narrow acquisition windows, and an issue with Adaptive RT having a jump in its estimated retention time. We have since done some work to mitigate this kind of occurance in 1.1, so we hope that the low rate of such occurence will be even lower now.

• Now if we press the Right and Left keys in Expert Review to look through the precursors with the most outliers, they look good. There are a few precursors that we would consider dropping from the assay, or at least from the Skyline file, due to low S/N or bad peak shapes, but the peak integration boundaries look nice.

Conclusion

Expert Review can work with label free data too, but the similarity of the reference data to the later experiments is important. Another thing that we could have done to improve the results using the GPF reference, is to use PRM Conductor on it that file, and then make sure that the experiment data use the same transitions as the reference. That would have been what one normally would do, but these data were used to demonstrate the point. Since Stellar is a PRM machine, using the new transitions would just be a matter of reimporting the raw files in the Skyline documents.