Thread by @Joni_Coleman

*Runs to bipolar* #WCPG2022

In the bipolar session, @MKoromina discussing fine mapping in bipolar GWAS. #WCPG2022

Pipeline: GWAS --> conditional analysis --> fine-mapping (PolyFun + SuSie, + FINEMAP) --> validation (eQTLs, colocalisation) #WCPG2022

Need to assess the effect of LD source and fine mapping windows. Can look across methods to ID consensus as probable variants to persue. #WCPG2022

ID a handful of small credible sets and a few SNPs of interest. Long term followers will be unsurprised to here that rs4702 [the best SNP] is a causal variant #WCPG2022

Characterised fine-map implicated genes - overlap with genes implicated in sequencing #WCPG2022

Most credible SNPs are the lead SNP, but not all #WCPG2022

Overlap of fine-mapping results with epigenomic peaks and eQTLs. Patchy, but most show some biological relevance from triangulating analyses #WCPG2022

Can leverage fine-mapping effect sizes to estimate polygenic risk scores (advanced form of beta shrinkage). Shows improvement over C+T, PRS-CS in R2 from comparison bipolar disorder cohorts #WCPG2022

What next? Incorporate in-sample LD for fine-mapping. Fine-map MHC. Improve PRS further with PolyPred (incorporates function). Fine-map in multi ancestry and admixed populations #WCPG2022

*Runs to eating disorders*

Back in eating disorders for @helena_davies_ talking about age of onset in eating disorders #WCPG2022

Each patient has a different trajectory, but a common aspect is the duration of untreated eating disorder, which often lasts years and has numerous negative consequences. Early identification and intervention would reduce these negative outcomes #WCPG2022

Previous studies of AoE in eating disorders show eating disorders typically begin in late adolescence / early adulthood #WCPG2022

Does this hold in UK samples (GLAD, EDGI, COPING)? Yes - AoE 18 years for binge eating, 17 for compensatory behaviours, 20 for low weight (but variance around these medians). More males report later onset #WCPG2022

Risk of patients falling between the gap of CAMHS and adult mental healthcare. Need to be cautious about generalisation, sample was young, White, female in UK. #WCPG2022

Genetics: previous analysis by @hunnawatson showed anthro rg was stronger in typical onset AN but not early onset #WCPG2022

Examined anthro and Psych multi-trait PRS on binge eating symptoms in GLAD and COPING. #WCPG2022

Split cases into onset < 18 years and > 18 years and compared to controls. PRS using @ollie_pain reference standardised PRS from GenoPred. Internal cross-validation to train parameters of model #WCPG2022

See different polygenic liabilities playing a role at different life stages. Prediction of early onset binge eating was better than late onset. Work ongoing and expanding - watch this space! #WCPG2022

Finally I'm staying put for @lisadinkler talking about ARFID in Swedish twins. #WCPG2022

ARFID debuted in DSM5 and ICD11 - not documented in older data, little research to date, need for validation of screeners etc. At the beginning of ARFID epidemiology. Need to approximate ARFID from other disorders, measures etc. #WCPG2022

What is the h2 of ARFID? Looked in Swedish CATSS study, examining self-reported eating behaviour and Swedish registries. Complex algorithm #WCPG2022

Anticipate ARFID being a stable phenotype - examined individuals 6-12 years. Could examine some aspects of ARFID diagnosis, but no evidence to rule out external limitations on food intake (availability, religion) #WCPG2022

Having ID'd children with restricted eating and/or evidence of eating disorders, excluding those with evidence of weight and shape concerns. Sensitivity analysis - exclude medical conditions, ASD.

667 cases, down to 539-541 in sensitivity analyses #WCPG2022

Point prevalence is 2%, sex ratio slightly female biased [I think... @lisadinkler did I get that right?] #WCPG2022

MZ correlations >> DZ - D? Sibling contrast effects? Also see indication of sex differences - evidence for qualitative sex differences, less so for qualitative. Prefer sibling contrast as explanation for now #WCPG2022

Best model is AE-s - A ~80%, E ~ 20%, s ~ -10%. ARFID above other EDs, similar to neurodevelopmental disorders - could be a partially neurodevelopmental disorder? Worth examining. Genetic studies seem worthwhile #WCPG2022

Now a panel discussion (we miss you, planned discussant @LauraHuckins!) #WCPG2022