The Skyline developers hosted a new webinar about their product and I spent the morning watching. While I learned a few things, I mostly realized how much I still have to learn on the subject of proteomics. Will start by watching the previous webinars, especially #14, where they go over batch scripting for the CLI Skyline for large number of samples. Will be useful.
What did I learn:
Most importantly, what demultiplexing means in regards to proteomics.
When setting up your mass spec isolation window scheme, you have the option of using the same isolation scheme between mass spec cycles, or having some form of offset, which gives you a staggered window setup between cycles 1 and 2 as shown below.
The demultiplexing happens when you compare some both cycles of some T+/-1, and one cycle of some T0. When you compare what happens in different overlapping sections, it allows you to effectively parse your T0 at a width half as wide as that measured (If your window is 24 m/z wide, you can parse 12 m/z using the demultiplexing). I still have to try to fully wrap my brain around how it works, but I *think* this is an effective way to visualize it at a very high level.
There was also a lot of information on assessing model fit and believability, and how there’s no replacement for actually looking at the chromatograms once you’ve winnowed your number of proteins down to something manageable. Machine learning is getting better, but it’s still not a replacement for human eyeballs and pattern matching ability.
All in all, I think this Webinar was at a much more advanced level than I’m currently operating at, but I picked up a few useful things, and also found out that they make available some well curated and understood data to experiment on, which is something I look forward to playing with. The data for Webinar 15 will be posted here soon. Webinar 14 data, which I believe will be similar, is available here. Oh boy, new things to play with!