Search results

Thanks, will try to take a closer look at this. So it's like they used all possible genes but weighed them by how much the SNP signal from the GWAS points to them? That would make more sense and make there results more interesting.

I tend to agree. For most of the loci there are multiple potential genes implicated and each of the genes are involved in multiple pathways. You could perhaps argue that there are more genes involved in the immune and nervous system than expected. But it's hard to say how many immune-related...

Yes look like you're right. It seems that they highlight the GLMNET lasso model because it did well on the test set (while selection of the model should have happened before that).

I think platelet-derived cfRNA was decreased in ME/CFS. EDIT: so that would mean that the platelets are less likely to rupture in ME/CFS?

The model that reached the best fit (in the 30% test samples) was GLMNET Lasso. It reached a maximum accuracy of 77% (84% sensitivity, and 68% specificity, testing AUC-ROC: 0.81 (95% CI: [0.694 to 0.926]). This is similar to BiomapAI which reported an accuracy of 72.5% and AUC of 0.82 (but...

They also found that in their machine learning models, the accuracy was highly variable depending on the predictors that were included. The main reason for this was thee fraction of platelet-derived cfRNA, which was apparently lower in ME/CFS patients compared to controls. Not sure what this...

They tested RNA outside of the cell which represents a mixture of RNA fragments released from many different cell types all over the body. Perhaps one of the more interesting findings is that they found no differences in Viral RNA Signatures between ME/CFS patients and controls. In the theory...

I suppose they did all that selecting in the 70% training set and then tested it in the 30% data that the model hadn't seen yet.

They give the example of HMGCR, an enzyme involved in cholesterol synthesis. Variants in the HMGCR gene identified by GWAS have only a small effect (I've read SNPs had an OR ≥ 0.9). However, drugs such as statins, which inhibit HMGCR, produce large reductions in cholesterol land cardiovascular...

RS stands for the relative probability of success (expressed relative to drug targets without genetic support). The authors extracted data on drug development from Citeline Pharmaprojects for monotherapy programmes added since 2000. The interesting figure is 1D. showing that RS is not...

Chris Ponting pointed to this study following the publication of DecodeME to argue that small effect size in SNPs do not indicate what the potential effect of a drug would be...

I'd also be interested in the full text if someone has acces.

I've been trying to understand this through a made-up analogy. Thought I might share it to see if it holds and if others find it useful or not. Suppose an illness is caused by a structure somewhere in the body that lets cells through that it should hold back, like a dam that is breaking. There...

It's not really important so no problem if they don't find the time to look at this. It's probably because I don't have any experience with GWAS that I don't fully understand.

This might warrant its own thread because it has a lot of info and advice that is not available in other formats. I made this thread about it on social media: 1) There's a new document with care advice for patients with (very) severe ME/CFS. It was published by the Austrian Society for...

The DecodeME participants were 85% females themselves. I think it's the default of GWAS because sex chromosomes have additional difficulties to analyze see: eXclusionarY: 10 years later, where are the sex chromosomes in GWASs? - PMC EDIT: added quote from the paper above

I kind of picked examples with high heritability though. There are other diseases with similarly low heritability estimates such as rheumatoid arthritis so the estimate for ME/CFS is nothing weird. But perhaps a bit lower than some expected.

I think it was this study: Repurposing large health insurance claims data to estimate genetic and environmental contributions in 560 phenotypes (2019) Lakhani et al. | Science for ME Here's the overview from Dibble et al.

To clarify: the link to the Neal lab data includes a search function where you can find the heritability estimate for other diseases using UKB methods similar to what DecodeME used. It seems that ME/CFS isn't like schizophrenia, Crohn's disease or diabetes type 1 for example, which have a...

I got the impression that the ME/CFS estimate still seems quite low compared to other diseases using the same method https://nealelab.github.io/UKBB_ldsc/h2_browser.html https://pmc.ncbi.nlm.nih.gov/articles/PMC3469463/