Discussion in 'Fibromyalgia and Connective Tissue Disorders' started by Andy, Jul 23, 2019.
Open access at https://www.ebiomedicine.com/article/S2352-3964(19)30473-6/fulltext
Care to elaborate @arewenearlythereyet?
The study looks to have been written up nicely - I mean they seem to have explained what they did quite well, which is a good start.
This was interesting:
So, they chucked out all the results produced from one hospital because they were affected by autoclaving. It makes me wonder how often metabolomics studies are affected by autoclaving and the results get used anyway.
This was nice I thought:
So, there were over 1000 metabolic features identified that were different in more than 30% of the data and that were consistent between the different hospital cohorts (that's each of the dots in the left hand graph). Of those, 228 features were different between patients and controls (i.e. statistically different and a big fold change - shown as green dots.). I'd like to see other metabolomic researchers showing their data in graphs like that one above left.
But they didn't seem to be able to identify many of these features (which may mean that they are missing lots of clues?).
I don't understand this next statement - how did they get down from 88 to 7?
The seven metabolites that were identified as different between fibromyalgia patients and controls are shown in the right hand graph above.
Damn, I need a break, so tapping out.
Here's a link to another recent fibromyalgia microbiome study.
There's more than one procedure done at the hospital that can alter metabolomics-data, not just autoclaving. And I wonder too how often it is not properly handled when analysing the data, although it is not always easy to see, so I guess you can only do your best. I'm also a fan of volcano plots.
To me it seems like the 7 they end up with were the significantly different features whereas the 228 was all different features (after the filtrering steps).
Edit: A bit unsure if it's the seven most significant features or seven they identified beyond just "tentatively" using MS/MS (basicly just sending a sample through two mass-spectrometers)? In the graph it says "identified" and in the text they talk about significance, but the graph are calling all 228 significant.
I would say the use of the phrase gut brain axis was the first red flag.
Then followed by poor methodology (faeces sample linked to serum metabolites) using a small sample
With no established reference for what things should be and low precision by only looking at genus
Then a rather bullish and misleadingly overstated conclusion on many many fronts.
I find this sloppy and extremely poor science I don’t have a more extreme tripe picture and I think a pair of bollocks would probably be moderated as offensive.
I am sure many others will find this interesting...leaky gut and gluten free brigade probably. They seem to love correlation studies with low power and overstated conclusions.
I haven't had the energy to read the paper properly yet...but realising I'm perhaps one of the few who'd recognise tripe when they see it made me feel very old!
Amy Proal on Twitter has some a thread with some comments on this study
Separate names with a comma.