Thursday, March 9, 2017
MetaMorpheus -- Amazing software and stategy for finding all the PTMs!
I'm gong to end my blog hiatus with the best paper I read while I was out recovering -- and it's this new one out of Wisconsin!
Let's start with a minor criticism -- if you saw the title of this article in your Twitter feed you might think that this is a review on the topic of PTMs and just go right past it. And you shouldn't pass by this one.
Here is the thing -- our database tools are really good at finding peptides from unmodified proteins in our databases. If your job as a proteomics scientist is to identify peptides from model organisms with perfectly complete annotated UniProt proteins that are not regulated in anyway by PTMs you are in the clear -- we've got all the tools for you. If, however, you are studying something that actually exists in nature (i.e., modifies virtually all of it's proteins with chemical modification combinations of some kind) it's still tough in this field. Our tools are designed for unmodified proteins. Looking for any modification is possible -- but computationally super expensive (example).
I LOVE this paper, btw. I had worried that my enthusiasm for it had something to do with all the painkillers from my knee procedures, but -- narcotic free -- still love the paper! Here is the idea.
1) Screen the data at a high level with a great big mass tolerance window and look for PTMs
2) If finding evidence of the PTMs -- take a FASTA and build a more intelligent FASTA (at this point it must be XML) that includes this stuff (think of it like a ProSight Flat file where instead of using biologically curated data to build your database you are building your database with PTMs on the fly with the data that you have in hand)
3) With your smart database research your RAW data with your normal tight tolerances so you get everything right.
If you're thinking -- "hey, I can do that, I have all the tools necessary on my desktop right now." You might be right. You can do a wide tolerance mass search, find all your deltaM masses, convert them to PTMs, make a better database (okay...maybe you can do that...I can't...unless now I'm firing up ProSight....building a Flat file and doing the rest of it that way....) and then research my database.
My response -- can you download a free piece of software right now -- that'll just do the whole darned thing for you? It's called MetaMorpheus and you can get it right now -- right here!
Okay -- so this doesn't come without a hitch -- you are STILL doing a huge delta M search to start your program -- and even as fast as Morpheus is the search space is tragically large. For one of their human cancer digests it takes 13 days to run the project on what sounds like a seriously beefy PC...but...to really truly get to the bottom of these PTMs with ultra high confidence of their presence and their site specificity -- in one workflow...??!? I can't wait to give this a try!!