Thursday, October 17, 2019

Need more power out of that protein quan output? Reinterpret with MSQRob!


Have you got a beautiful output out of your favorite software with thousands of quantified proteins but you're still at this point?


Do I ever have amazing news for you! What if you could just take that quantitative output (for real -- your MaxQuant or mzTab output CSVs) and reinterpret that quantification side of it with a super easy, shockingly powerful. inference tool? Maybe that's what you need to push you toward that biological interpretation!

Introducing MSQRob!

BOO? You already knew about this? Cause it's been out for 2 years?

Okay -- well, I didn't and I literally love it.

Edit: You can download it here.

Also Edit 6 months later for why you should use MSqRob and it's this literal quote: " an innovative solution to this key issue by introducing a hurdle model, which is a mixture between a binomial peptide count and a peptide intensity-based model component. It enables dramatically enhanced quantification of proteins with many missing values without having to resort to harmful assumptions for missingness"

First off -- if you're all fancy and smart and stuff, there is an R package for it and it's got an amazing walkthrough. HOWEVER --  it has a Visual Basic GUI thing that launches a Shiny App so you don't have to do anything at all.  It installs it right on your PC.


...which...well......I mean....since I've got a choice, I'm 100% certain I'm doing it this way.

Why is it better? Well, I've got 2 or 3 papers open right now and I'm confused which one is which, but it has something to do with a razorback gorilla ridge reduction or something. Its definitely some fancy stats stuff that I wouldn't be able to do otherwise.

In one of the papers they go to a ground truth dataset (the E.coli/HeLa spike in Fusion files from Qu lab at Buffalo) and it rocks the comparisons versus everything else they compare it to.

My favorite part is that this is a bonus. I'm using it as an additional interpretation on top of a dataset that I've already processed. It doesn't cost me anything but the time to label my columns appropriately.

Also -- it's compatible with everything. Lazy people like me can run it in Windows with the cool Shiny thing. Or you can use Linux AND it can be scaled up on a cluster.

And I stumbled into a goldmine of Rob Lowe/Chris Traeger gifs and images putting this together, and it's been literally impossible to pick my favorites, but this is pretty great....


No comments:

Post a Comment