Search results

This 2020 paper on the H-MAGMA method also includes a list of 1,841 pleiotropic genes that were found in 4 or more out of 5 tested neuropsychiatric conditions such as ADHD, autism, schizophrenia, bipolar disorder and depression. I checked the FLAMES list from above to see of any appear in it...

With covariate I meant a data column with the average gene expression in the dataset that you can choose to condition on or not. From what I remember, the Duncan et al. file only gives the relative gene expression in the dataset per cluster, so not the absolute gene expression for each cluster...

Perhaps the file provided by Duncan et al. 2025 paper might be useful? it gives a specificity score matrix file for Siletti et al 2023 atlas. EDIT: so it's each cluster's gene expression relative to the total gene expression of the dataset. It doesn't work with a covariate...

I noticed that the 2019 Watanabe paper on cell type analysis using MAGMA has a supplementary note where they explain their choice of using the average gene expression per dataset as a covariate, rather than just the absolute gene expression per cell type. It's basically what we already saw...

Ah apologies if I misunderstood!

Thanks but I think this is a step in the wrong direction. It mainly argues that treatments like CBT shouldn't be tested in randomised controlled trials.

I contacted the authors and they clarified that FLAMES requires the effective sample size ("the sample size is used for fine-mapping and should ideally be Neff"). So I'll try to do it again with the UK biobank controls and effective sample size.

Here's the cell type analysis for the 461 cell types from the Siletti et al. brain atlas (same pipeline as Duncan et al. 2025). The data below shows the European MVP subgroup only (this dataset includes a column imputation quality so I filtered r2 > 0.6). The one for the MVP-meta analysis is...

So I filtered out rare variants (MAF > 0.01). I also don’t want signals that came from only one ethnicity group only, as this might more likely reflect bias. The meta-analysis hasa column called ‘direction’ that indicates if the signal was found in the three substudies. So I required this and...

There are a lot of significant hits but I don’t think these are reliable because they are dominated by rare variants. You can see this in the QQ-plot, that the rare variants show genomic inflation, probably because they depend on only a small number of cases.

MVP-META I’ll focus on the meta-analysis first. The summary results can be downloaded here on the GWAS catalog: https://www.ebi.ac.uk/gwas/studies/GCST90480593 And there are some visualizations available using pheweb here: https://phenomics.va.ornl.gov/pheweb/gia/meta/pheno/Phe_798_1

There were almost 5000 of these ME/CFS-like cases in the MVP database which is divided into different ethnicity groups: African, Admixed American and European. The numbers are given below (taken from the supplementary material of Verma et al. 2024). EthnicityCasesControlsTotal...

The data on ME/CFS is based on electronic health records using PheCode 798.1. This not only maps to G93.3 in the ICD-10 CM but also a chronic fatigue unspecified (R53.82). So it’s likely a broader group than just ME/CFS. https://phenomics.va.ornl.gov/phecodemap/

I had a closer look at the genetic data on ME/CFS in the Million Veterans Program (MVP). It includes data on more than half a million US veterans. Around 90% are male with a mean age of 62, so a very different cohort to DecodeME. The main results are reported in Verma et al. 2024...

It looks like this now, with many regions shifted upwards. The Upper_layer_intratelencephalic no longer stands out. They are now on par with (e)MSN.

Great, thanks. So the results are largely comparably when using absolute gene expression? Think it was worth checking so that we can be more confident about the results. How did you avoid adding the dissection average gene expression as covariate in MAGMA: is there a option for that, that you...

I made this quick overview of case definitions for 2-day CPET studies in ME/CFS. Quite a few, including the one by Davenport's group used Fukuda criteria as inclusion. The studies by Van Campen/Visser mention that both ICC and Fukuda were assessed and it isn't very clear if Fukuda alone was...

Interesting, thanks for posting this. Based on genetic evidence on ME/CFS, I had the impression that the illness might involve altered cortico-striatal communication, as this has been found repeatedly in studies on fatigue and sickness behavior. Most of these brain studies have very small...

For what it's worth, I made these plots of the MAGMA analyses that have been done using the FUMA version of the Siletti brain atlas and ME/CFS data. This one is the data that forestglip posted using DecodeME. I ordered the results according to the 31 superclusters. It looks similar to the...

I suspect it's because without a covariate, pretty much all the signal will be for genes that are common in every cell, making it harder to notice the signals for genes that have more specific functions. It could be that it then picks up cells that don't have much specificity and mainly have...