This is off the proteomics topic completely! Here is the thing. I have a very skeptical friend with some ridiculously cool samples. Like -- if anyone else in the world is brave enough to get these things -- I don't know who they are. And -- we talked about doing metabolomics on these samples. But before I could have these samples I needed to be able to prove that I could...well...do metabolomics.... And -- I may talk a LOT -- but I'm not gonna pretend I know how to do something especially at the risk of wasting precious samples! I'll just spend a year of my spare time learning how to do it!
To learn the field, I did what I normally do -- I started a metabolomics blog -- and forced myself over the last year to read as many papers as I could on the topic. It is still a new field to me, so I know I've got tons to learn, but I may still link it to the right somewhere. Maybe someone will learn something from it, and I don't mind feeling dumb.
Okay -- so you can read a lot about something and that's awesome, but you need to run the instrument, clog a few columns and lose your temper with the software a few times to learn a new discipline, right? And it helps if you have some motivation....
...okay...so here is a perfectly anonymized map of region with 15 commercial vineyards within a 20 minute or so drive of one another. My favorite wines in the world comes from this area -- amazing and in my happy range of $8-$15 a bottle even after they arrive here!
I went to such efforts to anonymize this -- cause there is an exception here. I don't like one of these vineyards. They are using the same grape varietal as everyone else. They are using the same strict rules of their appellation, in terms of how long they have to age, etc., but there is something that I really don't like about what they make.
Wine is just a mix of small molecules, right? As good of an excuse as any -- and with wine you're not exactly material limited!
Over an undisclosed time period I collected 1mL from a number of different bottles of wine from this region. The rest of each bottle was disposed of in a manner that meets the strict ethical guidelines of my undergraduate fraternity. Once a number of samples were collected, I borrowed a Q Exactive classic system with RSLC3000 from an old colleague using some vague statements and a promise to clean the S-lens later (which I totally did).
There aren't a bunch of "Q Exactive wine metabolomics app notes" but if you erase the word "wine" you're set -- I found 2 that were very similar, couldn't decide which one was better (now know this one is -- warning .PDF download) and ended up on the following methodology (used the columns and flow-rates they describe, btw - oh, and I injected 5uL of wine on column, cause why not...?)
Metabolomics is, however, ahead of us (in my humble opinion) in terms of the data processing in some ways. In most of the software I've tried so far they start with what is quantified -- and statistically significant between their sample sets -- THEN they care about finding out what it is. They have massive reductions in their search space by going to the XIC and throwing out all the stuff that is 1 to 1. Who cares about the molecules that aren't changing? Not me!
To find what is significant, metabolomics software relies heavily on statistical tools.
Check this out --
This is a shot from Compound Discoverer (which, btw, is super easy to learn if you are using Proteome Discoverer 2.0 or newer).
This is one of the first steps in analysis -- Volcano plots showing the fold change of your compounds on one scale and the P-value (!!!) on the other. You can just take your list of statistically(!!) significant changes that you find graphically and export them into a darned list! Out of thousands of compounds detected in these weekend runs -- there are about 200 that are 10x up or down regulated with a p-value cutoff of 0.05. Wish you could do something that easy in Proteome Discoverer to get to the bottom of what is interesting...? I hope I'll have good news for you soon!
Interesting notes -- there are thousands of soluble small molecules that will stick to a C-18 column and ionize in a bottle of red wine! What?!? Initially, I'm thinking "that is way too high" but you've got small molecules from the grapes -- from the yeast -- from the wood of the barrels and stems -- so it doesn't seem that crazy...
Also -- and this is funny -- wines from the same vineyards cluster together just using PCA. Want to start a wine counterfeiter busting business on the side with your Q Exactive (if it is yours to do what you please, of course) it is really easy to do. This is interesting to me cause anything you read on that stuff is done with big FTICRs -- and they -- and they're hungry helium habit aren't necessary -- you can do this with a benchtop system easy.
That big circle? Wines from one vineyard in particular are quite inexpensive and multiple years were available. They clustered really well together. Proof of the terroir myth, LOL?
So -- the big question -- what is it about wines from that one place that are different than the others? To find this I've got to do one of them volcano plot thingies with the wines compared.
I strengthened my cutoffs to narrow the list way down -- yeah -- I'm not screening 200 compounds -- but I have a few huge outliers...and a few are quite informative....but this ends up being one of my favorites.
Wow -- that one is kind of an ugly looking peak -- and a lot of the samples are virtually zero so you can't see a good comparison -- but I'm still gonna leave it here. Check out the numbers, though! We're looking at something that is upregulated like 200 fold over my control bottles!
If you've got high resolution MS/MS fragmentation mzCloud does a good job of identifying things. It is pretty strict, though. Low mass fragment ions are wobblier than you'd think without a lower mass negative calibrant than SDS. I took a significant hit by not adding an additional lower mass calibration ion...but ChemSpider had no problem making the ID of this massively upregulated molecular species.
It is called 3-Methyl-4-octanolide -- but we generally call it "whiskey lactone" -- cause it is a big part of the taste of whiskey. Long story short -- it is significantly higher in some oaks than others. In American oak it is super strong compared to other oaks in the world.
Now -- this may have absolutely nothing to do with why I don't like wine from that one vineyard -- but, of all the other vineyards there -- as far as I have been told in follow up emails -- only one uses American oak in their barrels....guess which one? It is likely just a funny coincidence, but it makes a good story.
I wasn't going out to really solve this -- I wanted to learn the techniques and learn the software and it's funny to me that I used a couple weekends and some cutting edge technology to tell what I think is an interesting story. I actually started writing this up to publish, but then I got lazy. The important part is that --- I got some cells from culture prepped from my friends I mentioned earlier and the data for our ASMS poster (its on the last day, if you want to see something I'm actually putting time into) convinced them that I could be trusted with the REALLY cool stuff.