Jump to content

Mendelian Randomization thread *and* causal modeling/interpretability thread

Recommended Posts

Mendelian Randomization

mrbase IS the site


(when you do MR-egger/IVW on genetics of CRP + LDL-C, it seems to remove much of the "bad" effects of both CRP/LDL-C)



not surprising b/c CRP doesn't seem to be super-central to inflammation's damage (it's downstream of other inflammatory proteins) and LDL-C seems to slightly inversely correlate with epigenetic age [see https://pubmed.ncbi.nlm.nih.gov/21325005/ ]

George Davey Smith has written A LOT -https://www.broadinstitute.org/videos/two-decades-or-150-years-mendelian-randomization

also god I didn't realize how much more interpretable the research has gotten over the past 5 years

the easiest example of collider bias: intelligence and conscientiousness being anti-correlated in a pool of equally qualified applicants to elite competitive universities (or equally competent people!)



Kejun Ying's 2024 paper used MR on 40k epigenetic variants... (still dont know what it was causal on yet)


More general causal interpretability


Edited by InquilineKea
Link to comment
Share on other sites

  • InquilineKea changed the title to Mendelian Randomization thread *and* causal modeling/interpretability thread

EWMR (from https://clinicalepigeneticsjournal.biomedcentral.com/articles/10.1186/s13148-023-01590-x ):


The EWMR analysis identifies three causally age-protective CpGs (cg24011261, cg05551922, cg05055782) proxied by one SNP (rs62259939) in the GPX1 gene. These CpGs are also found to be candidates for causal positive effects on the telomere length [see Additional file 4: Table S4A] while at the same time increasing the odds of basal cell carcinoma (e.g. , for cg24011261).

I really want to understand EWMR better (mostly b/c Kejun Ying is relevant) but I guess it's not the most urgent thing. There aren't many papers on it...



However, it is worth noting that the EWMR approach focuses on genetically controlled methylation sites that were reported to have consistent effects across different tissues [2, 27] providing further evidence that genetically controlled CpGs are more likely to have common functions across tissues compared to those with less direct genetic control. Moreover, blood is one of the top tissues that shares a significant number of CpGs and meQTLs with other tissues, including skin.

Some skin-specific CpGs may not be genetically controlled but rather influenced by environmental factors, for example pollution levels or UV radiation. In these cases, the EWMR procedure would not capture the effect of such CpGs, even if the meQTLs data is collected from skin samples



Edited by InquilineKea
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Create New...