Neurodevelopment Genes Encoding Olduvai Domains Link Myalgic Encephalomyelitis to Neuropsychiatric Disorders, 2025, Lidbury et al

I took the first three variants in Table 1 (Top ranked genes) and all three have known quality issues noted on the Broad Institute gnomAD browser/database.
At least all the variants for the genes that replicated in the other cohort show "Pass".

PTPRD
https://gnomad.broadinstitute.org/variant/9-8436361-A-G?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/9-8318231-A-G?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/9-8409888-C-G?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/9-8845429-G-A?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/9-8897215-C-T?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/9-8901739-A-G?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/9-9270379-G-T?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/9-9829690-G-C?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/9-9904274-G-A?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/9-10254793-T-G?dataset=gnomad_r2_1

CSMD3
https://gnomad.broadinstitute.org/variant/8-113650725-C-T?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/8-113617156-T-G?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/8-114359441-G-T?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/8-114399612-A-G?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/8-114406336-G-A?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/8-114418955-T-C?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/8-114436474-T-G?dataset=gnomad_r2_1

RAPGEF5
https://gnomad.broadinstitute.org/variant/7-22184167-G-A?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/7-22278040-A-T?dataset=gnomad_r2_1

DCC
https://gnomad.broadinstitute.org/variant/18-50924132-T-C?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/18-50369520-G-C?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/18-50567129-G-A?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/18-50597529-T-A?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/18-50618359-G-A?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/18-50622857-T-G?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/18-50622885-C-T?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/18-50623189-G-A?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/18-50668321-C-A?dataset=gnomad_r2_1

ALDH18A1
https://gnomad.broadinstitute.org/variant/10-97367511-C-A?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/10-97392993-T-C?dataset=gnomad_r2_1

GALNT16
https://gnomad.broadinstitute.org/variant/14-69809143-G-A?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/14-69734498-C-G?dataset=gnomad_r2_1

UNC79
https://gnomad.broadinstitute.org/variant/14-94120712-C-T?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/14-93902973-A-G?dataset=gnomad_r2_1

NCOA3
https://gnomad.broadinstitute.org/variant/20-46270379-G-A?dataset=gnomad_r2_1
https://gnomad.broadinstitute.org/variant/20-46215501-G-A?dataset=gnomad_r2_1
 
Last edited:
At least all the variants for the genes that replicated in the other cohort show "Pass".
That's a relief. I checked PTPRD in the UK Biobank data and nothing jumped out at me. Is it easy for you to do the other gene variants you list?
 
Last edited:
That's a relief. I checked PTPRD in the UK Biobank data and nothing jumped out at me. Is it easy for you to do the other gene variants you list?
Sure. I wasn't sure if it'd be better to do all variants at once or per gene. Here is all except PTPRD together, but let me know if you want to see them split by gene: Link

Plots:
PTPRD
CSMD3
RAPGEF5
DCC
ALDH18A1
GALNT16
UNC79
NCOA3

Edit: I'll just paste all the variants here for easy copying if necessary:
PTPRD
rs7854171
rs996924
rs3847293
rs2570961
rs7866753
rs10815990
rs12341573
rs1746813
rs16930522
rs2498611

CSMD3
rs7833307
rs4876478
rs17608734
rs4311682
rs4354335
rs7002354
rs2942852

RAPGEF5
rs1859806
rs11766861

DCC
rs11082992
rs1560521
rs11874663
rs4995148
rs7233997
rs9957443
rs16956110
rs16956114
rs9956477

ALDH18A1
rs11188397
rs3750700

GALNT16
rs1296214
rs1890939

UNC79
rs55882426
rs28385502

NCOA3
rs623953
rs6066395
 
Last edited:
Thank you both.

Here's the chart @forestglip posted for RAPGEF5:

Screenshot 2025-06-19 at 2.20.49 pm.png

I'm interested to try to understand what we are looking at here. Tell me if I am understanding things wrong.

You have taken a gene that this study has identified as useful in differentiating people with ME/CFS. You used the UK Biobank database to assess the prevalence of the SNPs making up that gene in the people in the database labelled with CFS versus the whole population in the database.

So, is the x axis the position of the SNPs on the gene? And each dot is an individual SNP. So, at any particular position, there might be several variants present in the UK Biobank population.

And the y axis (-log10 (pv)) is the p value, so a measure of whether the two samples are likely to be genuinely different in the prevalence of that particular SNP. A -log10(pv) of 4 is 0.0001 (i.e. 10^-4). So, there does look like something interesting going on around that 22,400 kb position. (?) What sort of p values would you want to see in order to get a bit excited?

And SNPs are labelled with rs numbers.
 
So, is the x axis the position of the SNPs on the gene? And each dot is an individual SNP. So, at any particular position, there might be several variants present in the UK Biobank population.
Yeah, the gene is most of the width of the plot, and the x-axis is the position of the SNP on the gene/chromosome. If I look at GeneCards for the location of RAPGEF5 using the GRCh37 assembly which Gene Atlas is using, it says it is located at chr7:22,157,854-22,396,773. The plot goes from 22,100,000 to 22,450,000.

I'm not sure about the interestingness of the cluster of SNPs or a good p-value cutoff.

I don't know too much about this subject, and am not sure what to look for on these pages.
 
Last edited:
Ask clinicians who've seen thousands of people with ME/CFS—they'll almost uniformly tell you that individuals with ADHD, autism spectrum traits, and hypermobility syndromes (whatever label you use) are notably overrepresented. Both, ADHD and hypermobility are also overrepresented in ASD.

All of these conditions seem to exist on spectrums and are very likely polygenic, with environmental stress (note that infections during pregnancy have always been a leading theory in ASD) playing a significant role in how they manifest.

The idea that ASD might predispose individuals to other neurological diseases—even if it's not yet being systematically studied—makes a lot of sense. (If you've seen the severe dysautonomias in severe and very severe ME/CFS, there’s no doubt this is a neurological disease.)

Take Parkinson’s, for example: recent studies suggest it's more common in people with ASD. And when you think about what we know about ASD—particularly on a biomolecular level but also from a sensory processing and nervous system regulation perspective—it’s not surprising that other neurological conditions might 'follow'.

Apart from all that, ME/CFS also shares many very obvious symptomatic overlaps with ASD that you don't find in many other neurologic conditions.

I'm on the spectrum myself, as are others in my family. And while "spectrum" is a useful concept, in practice it often gets diluted. It’s become trendy to be “on the spectrum,” but labeling people who are merely a bit socially awkward as autistic isn’t clinically useful, imo. Like ME/CFS, ASD would benefit enormously from better stratification—clearer subtypes defined by biological markers, and more precise definitions.

All in all, it would make a lot of sense to find shared genetic vulnerabilities in ASD and ME/CFS. I can't read the paper itself and it is probably way above me anyways, I am merely speaking from a clinical and first person experience perspective.
 
Last edited:
I'm interested to try to understand what we are looking at here.
On top of what @forestglip explained about the plot you need to manually check variants that are higher in the plot that you are interested in to see if the MAF is too low. Low MAF values are not filtered out by the plot that would be filtered in a research study analysis.

If you look at the results in table form you can sort the table by pv and then check the MAF.
 
Replication of these results was attempted via a GWAS on raw data from a US cohort, which confirmed shared significant associations with variation identified in the PTPRD, CSMD3, RAPGEF5, DCC, ALDH18A1, GALNT16, UNC79, and NCOA3 genes.
One of the genes that was significant in both cohorts, DCC, was also significant in depressive symptoms not associated with recent stress:

S4ME Thread: Disentangling nature and nurture: Exploring the genetic background of depressive symptoms in the absence of recent stress exposure using a GWAS approach, 2025, Erdelyi-Hamza et al
We included nearly 200,000 subjects reporting no stressful life events in the past two years with data on current depressive symptom severity.
64 SNPs with suggestive significance were identified, one SNP (rs60939828 p = 5.92 × 10−11), located in DCC survived Bonferroni correction. DCC (p = 4.16 × 10−10) was also among three genes significant in gene-level associations.

DCC on GeneCards:
This gene encodes a netrin 1 receptor. The transmembrane protein is a member of the immunoglobulin superfamily of cell adhesion molecules, and mediates axon guidance of neuronal growth cones towards sources of netrin 1 ligand. The cytoplasmic tail interacts with the tyrosine kinases Src and focal adhesion kinase (FAK, also known as PTK2) to mediate axon attraction. The protein partially localizes to lipid rafts, and induces apoptosis in the absence of ligand. The protein functions as a tumor suppressor, and is frequently mutated or downregulated in colorectal cancer and esophageal carcinoma.
 
Last edited:
Things I want to keep an eye on, after reading and discussing the paper with an AI to better understand. (AI quotes slightly edited for clarity and links to GeneCards added):

CSMD3 and PTPRD had many significant variants, and included some of the lowest p-values and largest effect sizes of the 108 significant variants/78 genes.
Gemini AI said:
  • A variant in CSMD3 (rs17608734) has a p-value of 2.11×10−22. This is an incredibly strong statistical signal, far exceeding the others.
  • Multiple variants in PTPRD (e.g., rs996924, rs3847293) also show extremely low p-values, reaching as low as 4.55×10−11.

Both of these were also replicated in the Nevada cohort (not sure how the replication worked. Did they compare the Nevada cohort to the same controls, or did they use the controls from the original Nevada study?):
Study said:
Several associations shared by both cohorts were successfully identified, namely (1) a cluster harboured in the genomic region encoding protein tyrosine phosphatase receptor type D (PTPRD) (Australian cohort, p-value = 2.21 × 10−6; Schlauch et al. cohort, p-value = 1.14 × 10−6); (2) a cluster of six markers harboured in the CUB and Sushi multiple domains 3 (CSMD3); [...]

One of the two gene sets they found were enriched using their significant genes was Regulation of Dendrite Development (GO:0050773). The significant genes from this study that are in that gene set are DBN1, PTPRD, DCC, and NTRK2. (Note PTPRD included here.)

(The other significant gene set was IPR004092: Mbt repeat.)

As Hutan also quoted from the study, PTPRD is associated with a lot of things:
associated with cellular processes including cell growth, differentiation, the mitotic cycle, and oncogenic transformation. Members of the PTP family also promote neurite growth and regulate neurons’ axon guidance. PTPRD is one of the most frequently inactivated genes across human cancers, including glioblastoma multiforme (GBM) [74]. It is well known that pre-synaptic PTPRD promotes the differentiation of glutamatergic synapses, and several studies link PTPRD genetic variation to psychiatric phenotypes such as schizophrenia, bipolar disorder and mood instability [75], obsessive–compulsive disorder [76,77], and weight gain with antipsychotic medication [78]. Variants harboured in PTRPD have also been associated with susceptibility to the de-development of neurofibrillary tangles [79].



ALDH18A1
There were a couple significant variants related to this gene. This is also one of the genes replicated in the Nevada cohort.

From Wikipedia:
Delta-1-pyrroline-5-carboxylate synthetase (P5CS) is an enzyme that in humans is encoded by the ALDH18A1 gene.[5][6] [...]

The encoded protein catalyzes the reduction of glutamate to delta1-pyrroline-5-carboxylate [P5C], a critical step in the de novo biosynthesis of proline, ornithine and arginine.

Mutations in this gene lead to hyperammonemia, hypoornithinemia, hypocitrullinemia, hypoargininemia and hypoprolinemia and may be associated with neurodegeneration, cataracts and connective tissue diseases. Alternatively spliced transcript variants, encoding different isoforms, have been described for this gene.[6] As reported by Bruno Reversade and colleagues, ALDH18A1 deficiency or dominant-negative mutations in P5CS in humans causes a progeroid disease known as De Barsy Syndrome.[7]

The study noted that the metabolite that this enzyme makes, pyrroline-5-carboxylate (P5C), was elevated in ME/CFS in another study:

Metabolic features of chronic fatigue syndrome, Naviaux et al, 2016 [Article] [S4ME]
Pyrroline-5-Carboxylate and Arginine Were Increased.
Pyrroline-5-carboxylic acid (P5C) was increased in both males and females with CFS (Tables 2 and 3). P5C production is a well-studied response to stress in plants (26) and mammals (27, 28). P5C can be produced by the stress-induced oxidation of proline and hydroxyproline from collagen turnover via the enzyme proline oxidase or from glutamate oxidation via P5C synthase (P5CS).

So significant mutations in this gene in two cohorts, plus the metabolite it makes was increased in another cohort.



Other interesting genes because of the type of variant they had were CD8B, ACR, ALPG, and RFPL4A:
Gemini AI said:
Variants with Functional Consequences: Some variants are more compelling because they directly alter a protein.
  • CD8B (rs4514875): This is a stop_retained variant, meaning it could affect where the protein is told to stop being built. Since CD8+ T-cells are a key part of the immune system, this is a very interesting lead.
  • ACR (rs5771002), ALPG (rs183793479), & RFPL4A (rs147984855): These are missense variants, meaning they change an amino acid in the protein sequence, which could alter its function.



These have already been posted, but just re-listing the other genes that were replicated in the Nevada cohort:

RAPGEF5
DCC (One of the four genes that matched the dendrite gene set above.)
GALNT16
UNC79
NCOA3

And here are all the significant genes from this study, added for searching in the future:
ACR
ADCY7
ADGRE5
ALDH18A1
ALG1
ALPG
APOLD1
ARSG
ATR
ATXN2
BMP8B
CADM4
CD8B
CDK5RAP1
CEP170B
CLEC18A
CREBBP
CSMD3
CYP4F8
DBN1
DCC
DNAH14
DNAJB1
DPY19L2
EEF2KMT
ENTPD2
ERO1B
EVPLL
FAM86B2
FCGBP
GALNT16
GGCT
GOLGA3
GOLGA6A
GPSM1
HMGCS2
HSPBP1
ITGAX
JAG1
KIAA0753
KRT24
L3MBTL3
LAMA2
LAMC3
LGALS9B
LOC100506990
LRRC37B
MBTD1
METTL3
MKI67
MUC20
NBPF1
NBPF10
NBPF15
NCOA3
NNT
NTRK2
NUTM1
PIP4P2
PTPRD
RAPGEF5
RECK
REEP3
REXO4
RFPL4A
RGPD4
RIPK2
RSPH10B
SFMBT1
SLC38A2
SNHG9
SORL1
SPATA31F1
SULT1B1
TMEM87B
UNC79
ZNF266
 
Last edited:
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.
 
Does anyone have an idea why genes for laminin subunits would come up?

How does laminin relate to immune and neurological function?

In my mind, it could be that abnormalities in genes like laminin (which have a structural function as far as I understand) could have subtle effects on the digestive system and the immune system that is present there, which then could influence the immune system elsewhere.

Another seemingly obvious link could be that these genes also play a role in neurodevelopment and subtle abnormalities could influence the function of the brain. That onset of ME/CFS has a peak around 15 years seems compatible with neurodevelopment playing a role.

If I search for laminin and neurodevelopment it says neurons deposit laminin alpha5 to stabilize synapses.

Edit: but here only LAMA2 and LAMC3 come up.
 
Last edited:
Laminins are part of capillary wall structure - including 'basement membrane'. Their variety suggests that they are also local signpost molecules used to make vessels in different tissues seem different to migrating white cells, either directly or indirectly. They might be crucial to what cells are allowed to cross into brain.
 
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 agree. I feel like with expression, metabolite, brain imaging findings etc, ~90% of the findings might easily end up being products of lifestyle: low physical activity, diet, medication, etc. We've got a handful of the really good stuff, the genetic causal studies, with another really big one coming soon. I think it makes sense to really focus on these - figure out what genes or pathways show up in multiple studies, and do deep dives into the most promising genes.

The other observational studies feel more like they should act as a second step for confirmation and digging deep into mechanisms after you've got genetic leads, like how it worked out with the metabolite pyrroline-5-carboxylate above. Maybe not so much for exploratory research.
 
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.

This has important implications for treatment, no?

It explains why MECFS patients can get rapid remission of symptoms from Abilify (and other dopamine system stabilizer drugs) and Encephalitis lethargica patients get rapid remission from levodopa, but in both cases, the effects generally only last for a couple of months before pooping out. Maybe the continued immune system dysfunction eventually wears down the “temporary” dopamine/synapse fix.

By contrast, immune treatments like cyclophosphamide, Daratumumab, and JAK inhibitors (limited anecdotes) for ME/CFS can last for years.
 
Back
Top Bottom