What would successful brainstorming about ME/CFS genes look like?

Sasha

Sasha

Senior Member (Voting Rights)
From other threads:

Jonathan Edwards said:
At some point in the fairly near future I hope it will be useful to have a serious brainstorm on all the risk gene data available. I would limit that to risk gens rather than expression studies at first although the expression studied are obviously of interest to any ideas from the genes.

I still think we are going to see genes for adaptive immunity and synapses looking most interesting. The question arises as to whether these might underlying two separate processes - one each - or be in series. If in series I think we have a priority reasons for thinking immune response affects brain rather than brain affecting immune response but it would be good to have the arguments for that explicit.
Click to expand...

Sasha said:
... what do you mean by gene identification, and is that a process that the unskilled can help with? Is it a matter of scanning papers and looking stuff up?
Click to expand...

Jonathan Edwards said:
Anyone can have a good but I am thinking chiefly of the high level expertise we have here for making sense of the gobbledygook of genetics results, noting pitfalls and putting things in a tight perspective.
Click to expand...

Bears of little biological brain might struggle to contribute, but as new genetics papers get published and the experts gather to brainstorm, I'd at least like to understand the game, as I'm watching it while munching on my popcorn. What will the experts be looking for? How will we be able to tell if it looks as though they're winning? What scores points? For example, is this a rule?:

jnmaciuch said:
No genetics study can point to involvement of any system definitively unless those genes are exclusively expressed in one cell type
Click to expand...

And are there other rules?
 
In general, genetics studies give hints towards candidates that might be useful to test in perturbation studies/functional assays, and serve as fodder for theorizing about what cell types and pathways might be involved. The crucial part is just not getting carried away with conclusions and claiming that those results show ME/CFS must be an XYZ disease. Anything derived from a genetics study is still at the level of a narrative projection, subject to confirmation bias and all the failings of the human brain therein.

Even if a study points to one specific gene or cell type very strongly, a genetics study cannot differentiate between genes that are involved in maintaining the disease state and ones that make one more likely to end up in the disease state via a confluence of triggers, with any degree of indirect involvement. It can be helpful to look up top genes and see in what contexts they have been previously discussed—and that’s probably what many researchers will start by doing. But that can also lead people down rabbit holes of bias, since the literature itself is also biased by what people have chosen to study previously.

As I’ve said on other threads, you might find a top hit with the name “Very Important Gene For Neural Synapses” and it turns out that the main role of the gene in the disease is its role in electrolyte regulation in the kidneys. It bears remembering that the vast majority of genes are only understood in a fraction of their relevant roles.

I suspect that when new results come out, people will be inclined to align those results with their pet theories, as is human nature for us all. In all honestly, completely new insights are hard to derive from genetic studies because we simply won’t be able to comprehend the relevant connections if it isn’t already in the context of well-characterized biological pathways. Researchers will get excited that these results support what they suspected all along, they will do some functional studies, the majority of which will probably be duds, and then they’ll revise the theory and keep cracking on.
 
I think anyone can play this game. You just need to be able to search the internet, ask questions to AI, read up on Wikipedia and GeneCards, and see if any of that links to ME/CFS data or hypotheses in papers. And most of all, don't worry about what you don't know. Just like we would do for any new paper.

One must take into account that for many genes you will find information on cancer and heart as those have been studied the most.

I wish I could find the link to a recent paper on myasthenia gravis. I was absolutely stunned that now they have found so many gene defects and biological processes that cause the same outcome. Perhaps if researchers had thought more out of the box they could have found that information sooner. Maybe there are multiple clues in ME/CFS that will point to a particular pathway if only we could figure out how they are linked.
 
jnmaciuch said:
Even if a study points to one specific gene or cell type very strongly, a genetics study cannot differentiate between genes that are involved in maintaining the disease state and ones that make one more likely to end up in the disease state via a confluence of triggers, with any degree of indirect involvement.
Click to expand...
So, to use a comparison which may not help too many others, but a gene association may be like a function in a callstack of a crash dump? It may tell you something about a running process at a time of a crash but not necessarily the location or root cause, you need more information and analysis to get to that.
 
@forestglip provided a great example of looking up a gene and searching how it relates to published studies etc. I'm sure Dr. Karl Johan Tronstad would love to learn about the gene link to the M3 subtype in his study that was found. Here is the post.
forestglip

Post in thread 'Genome-Epigenome Interactions Associated with Myalgic Encephalomyelitis/Chronic Fatigue Syndrome, Herrera et al, 2018'

So the only significant finding in terms of genetic mutations was a SNP which affects the expression of CCK (aka cholecystokinin). The database they used seems to say that this SNP decreases expression of CCK in cultured fibroblasts but increases expression in colon cells. I couldn't figure out from the paper which allele (C or T) was the risk allele for ME/CFS.

This bit was very intriguing in the context of ME/CFS:
forestglip said:
Rats without the gene CCK experience less sickness behavior when administed bacterial endotoxins. (Source)
Click to expand...

Could it be that increased CCK plays a part in...
  • Like
 
I'd like to have a go at the process of checking out a gene, to see how far I can get and to get a better sense of what the bears of bigger biological brain will be doing.

I'm off to a great start, because I went to pick a gene from the Zhang paper and couldn't find a list of the 115 genes they identified as being associated! It all seems to be about groups of genes. Can anyone help me find a gene to play with?
 
Sasha said:
Can anyone help me find a gene to play with?
Click to expand...

We listed the genes on the Zhang thread somewhere.

It is worth having a go but remember that this is a bit like staring at a faded fragment of parchment and trying to work out whether the Bhagavad Gita was influenced by the Aryan Invasion. The Gene card may give you a rough date for the Invasion, and a rough radiocarbon date for that parchment but they may be wrong and the parchment is likely a copy anyway. Even with the data available you a bit like an archeologist staring at a single tooth and trying to decide what it means.

Having said that, some of us think this is a very useful exercise!
 
hotblack said:
It may tell you something about a running process at a time of a crash but not necessarily the location or root cause, you need more information and analysis to get to that.
Click to expand...

Except that the great thing about gene linkages is that they are always the root cause. It is just that they are part of the root cause and the root cause is the entire organism, its evolutionary history and its entire environment. As Leibniz pointed out the cause of an event is everything antecedent, including the infinite number of asteroids that didn't hit the earth the Thursday before (or the Friday).

I am more sanguine about the value of genes. In my past experience identifying genes like DR4, PTPN22 and B27 made a huge difference to our understanding of rheumatic disease. They don't tell you about detailed downstream mechanisms but when you already have a load of bits of jigsaw - including the blank blue sky bits - they can tell you which way up one lot of bits fits on to another lot of bits.

Moreover, the blank bits are really important - the things that aren't there in a given disease. They are almost more useful than the bits with marks on because they narrow down your options. ME/CFS is almost all blank blue sky bits which may actually make things easier, if finding a treatment only needs you to fit the marked bits together.
 
Aha! @forestglip had to ask the study authors for the full list of genes, since they're not in the original paper.

forestglip said:
The authors have sent me Supplementary Table 2 and allowed me to share it here. They also informed me that they've uploaded it (and I assume the other tables) to MedRxiv, but it'll take some time for it to be made visible. I attached the spreadsheet to this post which appears to have every gene they tested, but here are the top 115 that were used in the model:

Edit: Links to GeneCards added.
Click to expand...

Gene | p_value | q_value | Attention_difference_case_vs_control
DNMT3A | 9.827243889E-09 | 0.00002300778546 | 0.6657754794
ADCY10 | 0.0000004241863691 | 0.000496557788 | 0.3852176317
PPP2R2A | 0.000006694100941 | 0.005224131339 | 0.4087916389
NLGN2 | 0.00004209366361 | 0.00936668395 | 0.4034811495
LEP | 0.00005452940549 | 0.00936668395 | 0.3577475602
SYNGAP1 | 0.00005123040968 | 0.00936668395 | 0.343495285
AHCYL2 | 0.00002135782271 | 0.00936668395 | 0.3178829667
NLGN1 | 0.00006078993018 | 0.00936668395 | 0.3038524214
DLGAP4 | 0.00002541551244 | 0.00936668395 | 0.2603197173
HDAC1 | 0.00007201372681 | 0.00936668395 | 0.2325809437
AMPD2 | 0.00006720473825 | 0.00936668395 | 0.2268121061
AHCYL1 | 0.00005096416925 | 0.00936668395 | 0.2225109289
SHARPIN | 0.00005017326218 | 0.00936668395 | 0.2209758038
NME2 | 0.00003378467487 | 0.00936668395 | 0.2115389973
NME1-NME2 | 0.00006948111585 | 0.00936668395 | 0.2054587747
CACNA2D3 | 0.00005109712579 | 0.00936668395 | 0.1483053899
NME3 | 0.00006668949289 | 0.00936668395 | 0.1456878904
ZC3H13 | 0.00006720473825 | 0.00936668395 | 0.09569433479
CAMK2A | 0.00007675381042 | 0.009457784822 | 0.240018082
PIK3CA | 0.0000830396655 | 0.009720725517 | 0.3056303453
MAX | 0.00009539974472 | 0.01063582059 | 0.363732902
HLA-C | 0.0001033662833 | 0.01100016766 | 0.05538126826
ACE | 0.0001142039468 | 0.01162509102 | 0.2636674328
PRKCZ | 0.0001545910345 | 0.01204914093 | 0.4117194477
NFATC3 | 0.00014401371 | 0.01204914093 | 0.2631120614
DLGAP2 | 0.0001646883876 | 0.01204914093 | 0.2621262787
GRM1 | 0.0001371182639 | 0.01204914093 | 0.2185587328
RFK | 0.0001344477253 | 0.01204914093 | 0.1994060997
PELP1 | 0.0001534640069 | 0.01204914093 | 0.1929117364
NME1 | 0.0001587901151 | 0.01204914093 | 0.1852417535
AGO1 | 0.0001623048926 | 0.01204914093 | 0.1267421033
GDPD1 | 0.0001599537131 | 0.01204914093 | 0.1120424677
GRB2 | 0.0002021723225 | 0.01239980562 | 0.552413268
DLG2 | 0.0002136072243 | 0.01239980562 | 0.3455312025
NCBP2 | 0.0001854161799 | 0.01239980562 | 0.2924464065
CDC6 | 0.000217724158 | 0.01239980562 | 0.2719220561
COASY | 0.0002224441981 | 0.01239980562 | 0.1752380172
AK2 | 0.0001872117223 | 0.01239980562 | 0.1748823525
ENTPD8 | 0.0001805611026 | 0.01239980562 | 0.1734879474
PANK2 | 0.0002151424816 | 0.01239980562 | 0.12479525
PANK1 | 0.0002115761234 | 0.01239980562 | 0.1129610957
TOP1 | 0.0001936249297 | 0.01239980562 | 0.07420560482
HOMER2 | 0.000254026573 | 0.01292898413 | 0.2270032838
GABBR1 | 0.0002504529208 | 0.01292898413 | 0.1946995021
NAMPT | 0.0002475107416 | 0.01292898413 | 0.1810894201
NME4 | 0.0002446008164 | 0.01292898413 | 0.1620683964
GDPD3 | 0.0002644117404 | 0.01317121872 | 0.1161925477
NOTCH1 | 0.0002890727325 | 0.01327027847 | 0.3451436642
NRAS | 0.0002784390332 | 0.01327027847 | 0.3023653305
DNMT3B | 0.0002758381369 | 0.01327027847 | 0.2613961857
GNRH1 | 0.0002877231424 | 0.01327027847 | 0.2216789557
RBPJL | 0.0003043123279 | 0.0132863391 | 0.22416631
PRPF4B | 0.0003064474469 | 0.0132863391 | 0.1612924054
GALT | 0.0002993841048 | 0.0132863391 | 0.1392075018
BCL2L1 | 0.0003339523837 | 0.01370050545 | 0.1743575027
STAM2 | 0.0003394075499 | 0.01370050545 | 0.1526449203
TSC2 | 0.0003308718599 | 0.01370050545 | 0.1351486918
PSMB5 | 0.0003339523837 | 0.01370050545 | 0.09214654812
DLGAP3 | 0.000383478637 | 0.01496349385 | 0.255526225
PSMB7 | 0.0003817233905 | 0.01496349385 | 0.1038468629
IK | 0.0004069892243 | 0.01562054421 | 0.1592205242
ATP4B | 0.0004201960204 | 0.01586731086 | 0.1495895259
KRT5 | 0.0004447729991 | 0.01605659595 | 0.3562476643
DVL2 | 0.0004457832546 | 0.01605659595 | 0.2472893614
NR3C2 | 0.0004337993928 | 0.01605659595 | 0.1167869506
NODAL | 0.0004696144134 | 0.01665867928 | 0.1979832889
CDC23 | 0.0004792568214 | 0.01674698322 | 0.066298297
NEDD9 | 0.0005138908073 | 0.01694554671 | 0.3002752669
RET | 0.0005013445279 | 0.01694554671 | 0.170870038
CREB5 | 0.0005058736318 | 0.01694554671 | 0.07984582323
BNIP1 | 0.0005081523588 | 0.01694554671 | 0.07174378231
CA2 | 0.0005290926327 | 0.01720450978 | 0.09000343943
MICALL2 | 0.0005495999045 | 0.0175835427 | 0.2279392609
E2F6 | 0.0005557698915 | 0.0175835427 | 0.1783243055
PTPN11 | 0.0006335750884 | 0.01760595406 | 0.4921533247
PARD6B | 0.0005994539926 | 0.01760595406 | 0.2855943649
DLGAP1 | 0.0005836861957 | 0.01760595406 | 0.2772270668
HP | 0.0006349765183 | 0.01760595406 | 0.255350692
SMARCD3 | 0.0005954755686 | 0.01760595406 | 0.2501158964
PDYN | 0.0006391979969 | 0.01760595406 | 0.1762776835
STX10 | 0.0005915215514 | 0.01760595406 | 0.1228148241
SF3B2 | 0.0005928368537 | 0.01760595406 | 0.1120273812
NMRK2 | 0.0006238447535 | 0.01760595406 | 0.09930967091
RNF41 | 0.0006252263014 | 0.01760595406 | 0.08052012925
PSMB4 | 0.0006142525691 | 0.01760595406 | 0.07241156728
ENTPD5 | 0.0006888860687 | 0.01832769356 | 0.1475714803
AK3 | 0.0006828608689 | 0.01832769356 | 0.1425136165
PPCDC | 0.0006828608689 | 0.01832769356 | 0.1295292182
CHD8 | 0.0007010838189 | 0.01839838773 | 0.2717523155
CRKL | 0.0007651146253 | 0.01839838773 | 0.2465434884
ING3 | 0.0007292603404 | 0.01839838773 | 0.1827378544
CDR2 | 0.0007634502831 | 0.01839838773 | 0.159310581
PSMC3 | 0.000770127724 | 0.01839838773 | 0.1338064671
PSMC5 | 0.0007568262067 | 0.01839838773 | 0.1311539558
SCAF1 | 0.0007453613264 | 0.01839838773 | 0.1149503644
RAPGEF1 | 0.0007437366121 | 0.01839838773 | 0.07913227333
NT5C3B | 0.0007166110143 | 0.01839838773 | 0.05687172209
IL12A | 0.0007502551103 | 0.01839838773 | 0.04846638214
INS | 0.0007785501582 | 0.01841172533 | 0.3967527371
BUB3 | 0.0007904841128 | 0.01850700877 | 0.1242775519
HSF1 | 0.0008201711148 | 0.01882553742 | 0.09274033928
CHMP3 | 0.0008130950632 | 0.01882553742 | 0.07947528694
PSMD7 | 0.0008435614855 | 0.01899006663 | 0.1163095675
PDE4B | 0.0008399234731 | 0.01899006663 | 0.1089816052
CANT1 | 0.0008564089797 | 0.01909567427 | 0.1468378314
ADGRL2 | 0.0008807471749 | 0.01945308473 | 0.2623397449
BAIAP2 | 0.0009096190052 | 0.01946857239 | 0.2936694944
PPP2R2B | 0.0009057198106 | 0.01946857239 | 0.2153577511
ENTPD6 | 0.0009313416906 | 0.01946857239 | 0.1505787491
BHMT | 0.0008960389322 | 0.01946857239 | 0.1415637578
CREB3 | 0.0009253705382 | 0.01946857239 | 0.1162738525
PSMB3 | 0.0009253705382 | 0.01946857239 | 0.07935548515
CDC14A | 0.0009433903171 | 0.01954591724 | 0.1149938292
PANK3 | 0.0009535400255 | 0.01958290698 | 0.1051019582
SF1 | 0.0009741409195 | 0.01983202362 | 0.08197664579
 
So from that lot, I'm randomly picking the gene SYNGAP1, which the GeneCards link tells me is 'Synaptic Ras GTPase Activating Protein 1'.

The stuff at the top of the GeneCards page seems to be the guts of it and it says the stuff below. What do I do next (and how do I do it?)? (A slight drawback may be that I don't understand a word of the summary!)

Summaries for SYNGAP1 Gene​

NCBI Gene Summary for SYNGAP1 Gene​


  • This gene encodes a Ras GTPase activating protein that is a member of the N-methyl-D-aspartate receptor complex. The N-terminal domain of the protein contains a Ras-GAP domain, a pleckstrin homology domain, and a C2 domain that may be involved in binding of calcium and phospholipids. The C-terminal domain consists of a ten histidine repeat region, serine and tyrosine phosphorylation sites, and a T/SXV motif required for postsynaptic scaffold protein interaction. The encoded protein negatively regulates Ras, Rap and alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor trafficking to the postsynaptic membrane to regulate synaptic plasticity and neuronal homeostasis. Allelic variants of this gene are associated with intellectual disability and autism spectrum disorder. Alternative splicing results in multiple transcript variants. [provided by RefSeq, Nov 2016]

GeneCards Summary for SYNGAP1 Gene​

SYNGAP1 (Synaptic Ras GTPase Activating Protein 1) is a Protein Coding gene. Diseases associated with SYNGAP1 include Intellectual Developmental Disorder, Autosomal Dominant 5 and Syngap1-Related Intellectual Disability. Among its related pathways are MAPK family signaling cascades and Development Angiotensin activation of ERK. Gene Ontology (GO) annotations related to this gene include GTPase activator activity and SH3 domain binding. An important paralog of this gene is DAB2IP.

UniProtKB/Swiss-Prot Summary for SYNGAP1 Gene​

Major constituent of the PSD essential for postsynaptic signaling. Inhibitory regulator of the Ras-cAMP pathway. Member of the NMDAR signaling complex in excitatory synapses, it may play a role in NMDAR-dependent control of AMPAR potentiation, AMPAR membrane trafficking and synaptic plasticity. Regulates AMPAR-mediated miniature excitatory postsynaptic currents. Exhibits dual GTPase-activating specificity for Ras and Rap. May be involved in certain forms of brain injury, leading to long-term learning and memory deficits (By similarity). ( SYGP1_HUMAN,Q96PV0​
​
By using this website, you accept our Terms and Conditions.​
 
Sasha said:
So from that lot, I'm randomly picking the gene SYNGAP1, which the GeneCards link tells me is 'Synaptic Ras GTPase Activating Protein 1'.
Click to expand...

No you're not @Sasha. This is clearly a biased choice, based on your pet theory!!
Now you have started you can continue but you also need to pick another three genes that are complete gobbledygook and drive yourself nuts trying to figure those ones out.
 
Jonathan Edwards said:
Except that the great thing about gene linkages is that they are always the root cause. It is just that they are part of the root cause and the root cause is the entire organism, its evolutionary history and its entire environment.
Click to expand...
I suppose equally you could say that the entire root cause of a running program is it’s entire running code, memory and the environment itself running in (including most importantly every input received). A dump will show you the entire state at a given point, which sort of allows you to interrogate what’s going on, but it doesn’t capture everything about what has happened in the past (beyond understanding gleaned from seeing the effects of those things on the current state). It is a snapshot in time, which is one reason you may need more information (particularly longitudinal things like logs, other telemetry, etc) and analysis.

This may be an imperfect comparison but it sounds pretty similar to what you’re describing?

Either way I wasn’t trying to downplay the importance of genes, just get my head around things in a way more familiar to me I think.
 
Sasha said:
Also! Once I've got a gene, where do I look it up? Is it this GTEx database that @forestglip used, or GeneCards, that someone else used, are there others? Or is one enough?
Click to expand...
I'm learning just like you. But I think just read about it and then read about it and then read about it some more and then read about things related to it and hope a connection comes to you that might just possibly make sense in the context of ME/CFS.

I don't know how useful GTEx is. I just learned about it a couple days ago. I would start with reading the Wikipedia page of a gene (clicking links to other pages to better understand what they're saying), then maybe reading the linked references or searching on PubMed for papers about it.

And I don't know if you use AI, but I find it very helpful for putting advanced gobbledygook into easy to understand terms. I just paste a complicated sentence or paragraph and ask it to explain in simpler terms, and ask followups if necessary. It can still make mistakes, so be careful, but when it comes to explaining facts, as long as they're not very obscure facts, it's pretty good, as opposed to trying to get it to come up with new ideas/connections.
 
forestglip said:
And I don't know if you use AI, but I find it very helpful for putting advanced gobbledygook into easy to understand terms.
Click to expand...
I’d echo this. Stick to asking questions that are basically explaining a word or term, ideally after giving some context or source material, that way the models are usually regurgitating definitions from elsewhere. And if you ask about and prod the same topic from multiple angles over a week things will start to click a bit.

Back this up with using info from known good sources yourself and you should be okay.

Asking the models to think or reason or actually understand stuff or come up with original thought is where things often become much much more sketchy. But as summarisers/regurgitators they can be effective learning aids.
 
You must log in or register to reply here.
Back
Top Bottom