Thursday, July 28, 2016
SRMAtlas -- Tools to do SRMs on a bunch of human proteins!
Whoa! Finally got the full text.
I humbly present my interpretation of the HumanSRMAtlas that showed up in Cell last week.
Let me get this off my chest first. This thing has been around for at least 5 years, right? Seriously, I thought I'd been taking crazy pills, but then PubMed verified it:
With that out of the way! What is this?
Its a ton more SRM assays! They painstakingly developed synthetic peptides that would be representative of a large number of human proteins. They obtained MS/MS spectra for these synthetic peptides using some Q-TOFs and used multiple collision energies to assemble good spectra.
Then once they had this big library they picked a pathway that was interesting and they show that they can use triple quads to look at quantification of that pathway in human cells.
Will the SRMs still suffer from the same inherent problems of every low resolution MS/MS experiment ever done? Absolutely! But a lot of work went into minimizing this. Proof? Peptides from 7-20 amino acids long, they could identify with 96% success rate in their tissue (FDR ~4%) bigger peptides were more problematic (21-30 amino acids ~ 83% success rate), but remember this is SRM, you can always do more peptides.
Will interference be a bigger problem in more complex organisms like humans? Sure!
Will the quality of the data vary wildly depending on the body tissue? Of course.
But, this is a seriously comprehensive database. Loads of SRMs. A TON of work went into this and its a free resource out there at SRMAtlas.org!
Ben's biggest questions here -- presumably since this was designed with high resolution MS/MS, I wonder if this resource is PRM compatible for higher certainty, less noise, etc.,