jnmaciuch
Senior Member (Voting Rights)
My bad, I thought you were asking “and why” there were missing values
Multi-omics identifies lipid accumulation in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome cell lines: a case-control study, 2026, Missailidis et
- Thread starter Wyva
- Start date
Evergreen
Senior Member (Voting Rights)
Thank you for confirming that I can safely forget about this!Really hope you start feeling better soon.
jnmaciuch
Senior Member (Voting Rights)
Here’s the paper I was thinking of from Bali Pulendran’s lab about lipid regulation in B cells:
pmc.ncbi.nlm.nih.gov
It identified the transcription factor SREBP as the “master regulator” of B cell lipid metabolism—though other transcription factors could (and almost certainly do) also influence SREBP or subsets of its target genes [edit: and that is probably cell-type specific]
SREBP signaling is essential for effective B cell responses - PMC
Our previous study using systems vaccinology identified an association between the sterol regulatory binding protein (SREBP) pathway and humoral immune response to vaccination in humans. To investigate the role of SREBP signaling in modulating ...
pmc.ncbi.nlm.nih.gov
It identified the transcription factor SREBP as the “master regulator” of B cell lipid metabolism—though other transcription factors could (and almost certainly do) also influence SREBP or subsets of its target genes [edit: and that is probably cell-type specific]
Last edited:
This study is about underproduction of phosphatidylserine and BCR/CD22 and lymphoma.
Phosphatidylserine synthesis controls oncogenic B cell receptor signaling in B cell lymphoma
Phosphatidylserine synthesis controls oncogenic B cell receptor signaling in B cell lymphoma - PMC
Omi et al. identify phosphatidylserine (PS) synthesis as a key metabolic vulnerability in B cell receptor (BCR)-positive B cell lymphomas. Inhibition of PS synthesis causes an imbalanced phospholipid metabolism via membrane contact-based lipid ...
pmc.ncbi.nlm.nih.gov
I can't find a study about overproduction of phosphatidylserine and BCR/CD22, but I'll keep looking.
The principal role of CD22 is to regulate the BCR. CD22 is a membrane receptor found on B cells and is a member of the Siglec family of proteins ...
It does seem strange that the B cells of pwME/CFS had higher levels of phosphatidylserine than control group, but at the same time we have seen phosphatidylserine recommended for people with ME/CFS, and have seen people say they benefitted from it. Is that a clue?
Jonathan Edwards
Senior Member (Voting Rights)
I rather doubt it. I suspect that it is recommended simply because it sounds like a 'supplement' to alternative practitioners. And some people will always say they benefited from whatever.
From Linus Pauling website:
The long-chain omega-3 polyunsaturated fatty acids, eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), and the B-vitamin niacin (nicotinic acid) have potent triglyceride-lowering effects.
There is niacin in my B complex, but I have difficulty taking it separately.
@mariovitali, interesting that the pro-resolving mediators is based on EPA and DHA. I don't know if that's just a coincidence.
Maybe it is that triglycerides interfere with autophagy.
The long-chain omega-3 polyunsaturated fatty acids, eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), and the B-vitamin niacin (nicotinic acid) have potent triglyceride-lowering effects.
There is niacin in my B complex, but I have difficulty taking it separately.
@mariovitali, interesting that the pro-resolving mediators is based on EPA and DHA. I don't know if that's just a coincidence.
Maybe it is that triglycerides interfere with autophagy.
DMissa
Senior Member (Voting Rights)
We don't know if the observations seen here in a relatively small cohort of immortalised cell lines reflect a biological reality in the patient, and if they did what the relationship (if any) is with a disease process. So I would not endorse any particular intervention based on this study.
What I want to get funding to do next (with relation to this study) is related measurements on primary immune populations while coupled with relevant measurements of function that could plausibly be related to symptoms. But that's still just an idea in the oven and I am also weighing up its value against other avenues we are uncovering in other projects.
In case this was not apparent I am the first author on the paper
What I want to get funding to do next (with relation to this study) is related measurements on primary immune populations while coupled with relevant measurements of function that could plausibly be related to symptoms. But that's still just an idea in the oven and I am also weighing up its value against other avenues we are uncovering in other projects.
In case this was not apparent I am the first author on the paper
hotblack
Senior Member (Voting Rights)
Can you detail a bit more about what this would look like? A wider breakdown of B and T cell populations? Would it include information about innate immune cells too? When you say function do you mean function of the cells? What sort of measures would this include?
I’m just trying to understand what sort of information we could get from such a study.
V.R.T.
Senior Member (Voting Rights)
Exciting that you are uncovering other avenues! Are any of those projects written up/submitted or are they still ongoing?
@DMissa, I think you were going to come back to this?
The paper tells us that all of the samples were put through the mass spectrometer on the same day.
But then there was the liquid chromatography, and as far as I can tell, samples are put through one by one and there is certainly scope for the baseline to wander, even just with draughts and changes in humidity. How were the samples were ordered? Can you think of anything that might have been different between the ME/CFS and controls samples that could account for the remarkable result presented in Figure 3A?
How many times were the samples assessed? I'm assuming that if you had re-run the analysis for the lipid in Figure3A, to check things, you would have reported it. Is there any scope to redo the analysis, mostly just of that lipid? Maybe the analysis could be efficiently tacked on to someone else's study?
DMissa
Senior Member (Voting Rights)
Sorry I either forgot or didn't see.
Everything would have been done together. I don't think there'd be reason to worry about this. You won't get column degradation within same run that would confound things to the extent of seeing differences to that degree. As far as I know everything was run once. This stuff is all very standardised and run by these labs many times a week.
I'm sorry, I'm not following. Why would I have re-run this analysis? Why would it be repeated?
If I was aware of something it would have been in the paper or maybe not have even been submitted as it was. I am not aware of any confounders that would explain the differences. Definitely wouldn't be sidestepping around anything I was aware of
Thanks @DMissa. For sure I'm not suggesting that you are side stepping around anything you are aware of.
Thanks for the reply. I was reading about good practice for mass spectroscopy omics studies. It is recommended that samples are run again, focussing on the results of interest, if anything interesting is found in an exploratory analysis with thousands of molecules. There are just so many things that can affect results.
For the liquid chromatography stage, it's really important that the sample order doesn't add bias. As far as I can see from the manual for the machine that you reported was used, the samples are put through sequentially. Could you let us know the day and time for the samples? If the ME/CFS samples were spread out fairly evenly across the period of analysis, it's one less possible reason for a false result. I'm assuming that it isn't hard to check that, I think it is worth the effort.
It's just that that one result separates the controls from the ME/CFS sample in a way that we rarely see. For sure, the lymphoblasts are very manipulated and there are many things that can happen in the storage and analyses, and with the many molecules assessed an interesting result could be just down to chance. Maybe this result means nothing much. But we see so many historic preliminary studies that seem to find something interesting that then is never followed up on, so I'm keen that we get as much as possible out of clues that come up on our watch.
Also, many people in the community see preliminary studies like this and assume that this is evidence of the many biological differences found, when in fact it might not truly be a difference. And it sounds as though you may move on to other analyses, possibly based on these results. So, I'd just like to explore if there is a way to add more certainty (or less) to the result before you (and we) move on to other things.
Thanks for the reply. I was reading about good practice for mass spectroscopy omics studies. It is recommended that samples are run again, focussing on the results of interest, if anything interesting is found in an exploratory analysis with thousands of molecules. There are just so many things that can affect results.
For the liquid chromatography stage, it's really important that the sample order doesn't add bias. As far as I can see from the manual for the machine that you reported was used, the samples are put through sequentially. Could you let us know the day and time for the samples? If the ME/CFS samples were spread out fairly evenly across the period of analysis, it's one less possible reason for a false result. I'm assuming that it isn't hard to check that, I think it is worth the effort.
It's just that that one result separates the controls from the ME/CFS sample in a way that we rarely see. For sure, the lymphoblasts are very manipulated and there are many things that can happen in the storage and analyses, and with the many molecules assessed an interesting result could be just down to chance. Maybe this result means nothing much. But we see so many historic preliminary studies that seem to find something interesting that then is never followed up on, so I'm keen that we get as much as possible out of clues that come up on our watch.
Also, many people in the community see preliminary studies like this and assume that this is evidence of the many biological differences found, when in fact it might not truly be a difference. And it sounds as though you may move on to other analyses, possibly based on these results. So, I'd just like to explore if there is a way to add more certainty (or less) to the result before you (and we) move on to other things.
Last edited:
DMissa
Senior Member (Voting Rights)
There is no way that a systematic effect at the instrument level explains this particular observation. Almost all of the lipids are elevated in the ME/CFS LCLs. If there was a systematic effect due to sample order or whatever else it would certainly be associated with that broad increase. How then would it explain the reduction in that particular lipid?
If there was a way in which I could efficiently add certainty and interpretability to these results I would, but we're limited. It would be expensive to run this again to validate against itself and as I say this would be unusual and a less efficient use of resources than alternatives.
What we are planning for some future work is to run duplicate samples for each individual which would be an even better way of addressing this kind of thing. But that basically doubles the cost of your study if sample preparation is laborious (for cell culture it is) so it shouldn't be held as a standard to apply to everyone doing this stuff. It's a luxury we have for the particular project I am referring to.
I promise you that I am engaging in good faith but this is not a request that is going to be productive for anybody to follow up. It will not tell anybody anything useful and it will involve me writing to another lab to dig through their notebooks from something like 5 years ago. The mass spec vials will have been loaded and run at one time. That is guaranteed. It doesn't matter which day or time it is
I understand the drive to understand this better, but the the sheer number of (complex) possible explanations is why I am not going into this in more detail without more specific experimental follow-up. It's just too hard to interpret yet without making assumptions and leaps.
I have never encountered this being done. What one would normally do instead is run a targeted mass spec approach (as opposed to an untargeted mass spec approach as we undertook here), or instead do other validation assays. If people really care about a specific target, for proteomics they will do westerns to validate, and for RNAseq they will do qRT-PCRs. They don't run the mass spec or sequencing again. Lipids are also a bit different. They do not generally have as specific a functional role that a gene product or polar metabolite would have. So focusing in on a specific one with a subsequent replication run would at best validate that there's a difference but it still wouldn't be any easier to make biological sense of. It would be more productive to move on to more interpretable sample types and involve more detailed assays and then see if it pops up amidst everything else that we'd be measuring as I suggest in the next comment. Then, once we can make sense of it and we know whether it's replicating, we'd be better served by focusing on it.
I don't think I would be designing a study around that particular observation, specifically. If I did submit a funding proposal for follow up work it may, for example, be around assessing immune populations and functional measurements and associations with different modes of metabolism in primary cells. If that dramatic difference was some universal or B cell metabolic feature of the illness it would come up again and then it would be grounds to do something specific about, and it would be more interpretable with primary cells and good immunotyping. I don't see much value in running away too much with something that isn't replicated and that is impossible to interpret with only the current type of data (and sample type).
If there was a way in which I could efficiently add certainty and interpretability to these results I would, but we're limited. It would be expensive to run this again to validate against itself and as I say this would be unusual and a less efficient use of resources than alternatives.
What we are planning for some future work is to run duplicate samples for each individual which would be an even better way of addressing this kind of thing. But that basically doubles the cost of your study if sample preparation is laborious (for cell culture it is) so it shouldn't be held as a standard to apply to everyone doing this stuff. It's a luxury we have for the particular project I am referring to.
Last edited:
Thank you for the context around these lab processes, @DMissa.
Honestly, between the finding of general lipid increase, and that one specific lipid with virtually zero overlap, the latter seems more exciting to me.
I'm sure you probably have other opinions due to more experience and better knowledge of how the funding works, but it feels to me like it'd be more fruitful to build up from what may be something close to the core of ME/CFS (based on the near perfect split) to try to figure out how that can be interpreted, than to go from more interpretable findings, but where those findings less clearly split the groups, and thus are likely more auxiliary findings.
I hope none of this is coming across as criticism of you. Rather, I'm just excited about that specific lipid, and since false positives do happen, whether through chance, confounders, or artifacts (not saying this happened necessarily), I'm just thinking about that possibility to calibrate my expectation about that lipid.
I appreciate you continuing to explain about how all this works.
Honestly, between the finding of general lipid increase, and that one specific lipid with virtually zero overlap, the latter seems more exciting to me.
I'm sure you probably have other opinions due to more experience and better knowledge of how the funding works, but it feels to me like it'd be more fruitful to build up from what may be something close to the core of ME/CFS (based on the near perfect split) to try to figure out how that can be interpreted, than to go from more interpretable findings, but where those findings less clearly split the groups, and thus are likely more auxiliary findings.
I hope none of this is coming across as criticism of you. Rather, I'm just excited about that specific lipid, and since false positives do happen, whether through chance, confounders, or artifacts (not saying this happened necessarily), I'm just thinking about that possibility to calibrate my expectation about that lipid.
I appreciate you continuing to explain about how all this works.
DMissa
Senior Member (Voting Rights)
I don't necessarily disagree. All I am getting at is that this difference (or any of them) could be a downstream consequence of some other change in the pre-immortalised B cell populations that involves (or combined with immortalisation, leads to) a shift in metabolism. That may even be the most likely explanation. I would prioritise seeing confirmation that this is even relevant in primary cells first. That's the crux of it. I'm not saying I'm not interested in it, it just wouldn't be the sole focus of a proposal. Another issue is that it's a lipid and not a gene product. It's way harder to make sense of or put forward a compelling case about in isolation. And we can measure it alongside other things anyway. Such as doing more lipidomics paired with whatever other approach we adopt. It's not going to get sidelined or forgotten, it would just be checked alongside other things as part of broader projects.
Even if it did come across to me that way it's fine, I am not averse to criticism and don't think I am above it. But I am interpreting everything as being genuinely constructive so no need to worry or qualify things either way. I wouldn't be here on the forum pretty much every day if I didn't think the discussions here were leading a lot of the genesis of important ideas in the field. I do genuinely believe these discussions are easily among the most important being had in the world right now (for ME/CFS).
Of course, I only care about outcomes. We all need to share our ideas and critiques for that to happen. And researchers need to be receptive to the critiques. I have rebutted some of them but it's not to deflect or dismiss, it's based on genuine reasoning. I see the slippery fish "clarification" game all the time and I am not interested in doing that
Last edited:
Thanks DMissa. It's absolutely not a question of good faith. I had assumed that you had done or supervised the lab work yourself and would remember what happened, and that there would be computerised records of the analysis times for the samples relatively easily to hand as well. It sounds as though someone else did this analysis 5 years ago?
My understanding is that the liquid chromatography is done to separate out the various molecules in a column, one column per sample. The sample is forced under pressure and heat along a capillary, with the specific molecules travelling different distances and so lining up more or less separated from each other. And, then the columns are assessed with mass spectroscopy, with the MS able to do a better job now because the molecules are neatly lined up, ready for analysis. You and the paper clearly say that the mass spectroscopy was done as a batch, and in one day, and that's great.
But, and I'm floundering around in this for sure, but it looks to me as though samples are run through the liquid chromatography sequentially. (I'm learning a lot about what is important to watch out for, and certainly choosing the best techniques to isolate various sorts of molecules and specifically various sorts of lipids looks to be really complicated, with tradeoffs. Clearly, when we evaluate papers, we need to pay more attention to the specifics of the machines and the techniques than I have done before because they have a big impact on what is found and what isn't.)
The method in the paper mentions a run time of 30 minutes for the lipid analyses. You had 32 samples. That's a lot of hours in one day if the samples had to be run sequentially, and that's not counting time for sample preparation, calibration and other QC and cleaning. Maybe some samples were run in parallel? Think of this as helping me and maybe others be better evaluators of the next gas chromatography/mass spectroscopy paper.
As far as I can tell from the literature, it can matter a lot what day and what time of day the analyses are done, which is why typically so much attention is given to block randomisation.
As you say, if it is decided that the finding of that one markedly discriminating lipid is not worth replicating (or even just running a couple of the samples again to check it) because it is from highly manipulated lymphoblasts, then maybe the other findings in the paper don't have a useful level of reliability either?
chillier
Senior Member (Voting Rights)
When I was doing mass spec work the HPLC machines and Mass Spectrometers were run in a windowless room where the temperature and humidity was very tightly controlled. We would run 96 samples from a plate back-to-back in one non stop run. So at something like 30 minutes a sample the machine would be running for 48 hours. Once set up the whole thing was automated, the robot would pick up the samples from the plate and feed them into the column with the appropriate buffers and that would feed directly to the spray needle and into the mass spec. We could VPN into the mass spec's computer to check everything was running ok if we weren't in the lab.
I'm guessing @DMissa and co would have sent the samples they prepared to a mass spec facility and wouldn't have been directly running the machines themselves. The samples were probably ran in conditions very similar to what we used - the technicians ought to be aware of possible biases if the facility is any good.
It's a shame there probably isn't a simple biochemistry technique like ELISA that could be used to validate specifically that one significant lipid, because it wouldn't be able to discern it from all the other lipids in that class.
I'm personally more interested in the global changes in lipids than just that one specific one, because global changes in triglycerides and phosphatidylcholines has been reported quite a few times in ME/CFS. I also find it hard to understand why this specific lipid at that exact length with that exact number of double bonds (no higher or lower) should really be much more important than the others - but maybe there really is a reason.
I'm guessing @DMissa and co would have sent the samples they prepared to a mass spec facility and wouldn't have been directly running the machines themselves. The samples were probably ran in conditions very similar to what we used - the technicians ought to be aware of possible biases if the facility is any good.
It's a shame there probably isn't a simple biochemistry technique like ELISA that could be used to validate specifically that one significant lipid, because it wouldn't be able to discern it from all the other lipids in that class.
I'm personally more interested in the global changes in lipids than just that one specific one, because global changes in triglycerides and phosphatidylcholines has been reported quite a few times in ME/CFS. I also find it hard to understand why this specific lipid at that exact length with that exact number of double bonds (no higher or lower) should really be much more important than the others - but maybe there really is a reason.