Thursday, November 15, 2018

I'm sorry to say the 90s are over. Maybe it's time to move past SeQuest! the early 90s there was a SciFi TV show about submarines that was on network TV in the U.S., unlike the hairstyles of the actors, the show hasn't aged all that well.

SeQuest the search engine has faired a good bit better. Heck, I bet if you looked at every proteomics paper that has been published this year the majority used this awesome -- way ahead of it's time -- bit of awesome out of the Yates lab.

However, I think I'll always remember 2018 as the year of the PTMs that SeQuest got wrong...and...well the multiple times I crossed flood water far too deep to be safe, cause it won't stop raining on the U.S. East Coast -- but mostly all the darned PTMs I've had to manually evaluate and tell people SeQuest got wrong. I desperately wish I was done calling people, but I'm not...

On the UP SIDE! It's 2018!  We've got options! More than we've ever had.

I looked it up. According to Car & Driver, the best car in 1997 was the Audi A4. I got to drive one around that time and it was AWESOME. But by today's standards.... 21mpg highway (before the EPA here readjusted fuel economy, so -- probably 18 mpg highway) 0-60mph in 9.5 seconds. WHAT?? We were impressed by that? My car is old now, averages well over 45mg and is over 1 second faster -- and it's driven by a moron. All the time.

I'll always love pastel Reebok Pumps, 90s cars and SciFi, and SeQuest. For real. But I'm a grownup -- you probably are too -- and if we take a step back and detach our emotions we can see that all those things actually totally suck.

Convinced? No? Good! What if I show you more options we have? (Click to expand, if you want!)

This is how I've been running my searches today.

Please note. No JNCO Jeans. No SeQuest.

This is all stuff that is very close to launch (why else would someone let me have it?)

You know ProteomeTools, right? 300,000 synthetic human peptides? What if MSPepSearch was powerful enough to search libraries that big? Cause the new one coming totally will be! takes 5 seconds to search a RAW file. No joking. Okay...this is a small RAW file. But that's hella fast.

Oh. And you can Percolate it.

Then anything that doesn't match my spectral library? Send it to MSAmanda 2.0!

Okay -- I'm going to back up on my SeQuest criticism. If you have a lot of peptides per protein -- it's fine. Go with it. But if you HAVE to prove that peptide is real? For the love of dog, please do not use that engine!  MSAmanda is free and is so much better. If MSAmanda tells me a PTM is probably is. SeQuest? Break out the ruler. It's manual validation time....

Guess what? You can Percolate that too! (So now you're using equivalent metrics for both engines!)

Didn't match that? Okay -- well -- I don't know when this is coming out for everybody else, but I have a de novo search engine in PD!!!  And I can Percolate that, too!!  (More details later. I updated my PD and it broke my Novor node...and my Xcalibur...but now I have 4.2!

These are the tools for Proteome Discoverer fans, but if PD isn't your platform:
MSFragger is blowing up and the data I'm seeing from multiple groups says  >SeQuest
MetaMorpheus gets better every week and is way >SeQuest
Mascot is getting updated at least once/year and >SeQuest
PEAKs? MS-GF+, the 75 search engines in SearchGUI (like Comet!! which is >SeQuest), Byonic, on and on.
Oh and some MaxQuant/Andromeda/Perseus thing. I'm getting very re-familiar with now thanks to BoxCar -- and -- there are very good reasons that people are using it. And on and on...

Look, I'll probably queue up some mouse runs before I go to bed on SeQuest. And not just because mouse studies of human cancer when you've got a hospital full of humans with human cancer on your campus is stupid (cue a biologist to tell me things like -- "Ben, I can't test a new drug on humans"...or "...sigh....I need sections from brain for this study....I can't exactly take these from humans..sigh..." To these comments, I say please see the disclaimers section or the "driver of Ben's car is a moron statement above.")  😇🙉😇, but because I've got deep fractionation and I'm probably going to average 30% coverage of the proteins for all this work. SeQuest will be just fine. Will I get all the spectra? No. Will I be able to trust my PTMs? After you check them with a calculator and ruler, sure!

This is more of a rant that started with something about submarines, but it's a new age and we've got so many awesome tools to work with now!

1 comment:

  1. I'm super interested in that Novor node, but my google-fu isn't finding any code to import it into PD... help???