Thread by @Joni_Coleman

First session of #WCPG2022 proper! @jorsmo welcomes everyone back to meatspace, before introducing @BFranke_lab, one of the program committee chairs, who in turn will welcome everyone to the conference and it's theme of vanishing boundaries across multiple aspects of the field.

BF now introduces Prof Sarah Medland to give the first plenary of #WCPG2022, who will talk about the work of the ENIGMA imaging genetics consortium.

ENIGMA aimed to improve the reproducibility of imaging genetics. Apparent that although imaging costs more, still need the same scale of sample size as other GWAS. The solution - collaboration #WCPG2022

ENIGMA is now 12 years old, had thousands of investigators and 50+ working groups studying the genetics of human brain structures and function #WCPG2022

MRI for dummies - brain mostly water, water has magnetic protons, MRI uses radio waves to knock protons out of alignment, measure time for realignment. Time varies by tissue, so can use this to determine tissues. #WCPG2022

Data converted and stored as 3D image. A couple of important steps - spatial normalisation (aligning multiple samples) and segmentation (extracting regions of interest) #WCPG2022

ENIGMA is in part a story of methods challenges, with parallel journies of imaging and genetics. Both need careful quality control - blind combination --> nonsense #WCPG2022

Can measure phantom (reference fake brain) and show perturbation within and across scanners. Needed to overcome this with statistical manipulation #WCPG2022

Deidentification of individuals - genetic studies often do not release allele frequencies for this reason. Images have faces - need to deface, but humans and AI are able to re-face these images, so need to avoid raw data, do analyses on summary statistics #WCPG2022

Multiple testing - multiple levels of analysis, so flexibility in multiple testing and potentially very high testing load. Multiple atlases that can be used. Imaging space is avoidant of hypothesis free space because of this. Need to prioritise #WCPG2022

GWAS use 5*10-8 (although secondary analysis tends to use all variants). Imaging uses FDR - without this, get interpretable noise (e.g. dead salmon paper) #WCPG2022

ENIGMA - first paper had 22 cohorts, 10k individuals. Nothing significant in individual cohorts*. Meta-analysis had significant findings. One 206 author paper instead of 22 null papers

* APOE was, but it was Alzheimer's and they already knew it would be there #WCPG2022

Next ENIGMA had 50 cohorts - "if you build it, it will grow". Most hits structure specific. Subcortical regions significantly influenced by common variants, grouped by developmental, functional subdivisions. Effects similar to other complex traits - need big samples. #WCPG2022

Expanded sample sizes, got more hits. Looked beyond cortical regions into cortical thickness and surface area. 38k individuals, 15k replication. 34 regions of interest, 2 global traits, 369 GWS loci. Enrichment in Wnt signalling pathway, fitting established hypotheses #WCPG2022

Polygenicity looks similar to complex traits, 34% heritability, 3% from PRS. Strong positive correlation with cognition, Parkinson's, insomnia, depression, ADHD, driven by multiple regions #WCPG2022

Correction for total head size makes results orthogonal, forces no intercorrelation. Don't correct, see some intercorrelation [less than psych?] #WCPG2022

Strong region association in precentral SA (which also strongly differs between great apes and humans - ongoing work in this space). #WCPG2022

Because the universe has a sense of humour, the volume of the insula (of interest in depression) is associated with variants near SLC6A4 (popular candidate gene, definitely not associated with depression)... But the variants seem to be acting on EFACAB5, not SLC6A4. #WCPG2022

Educational attainment PRS are associated with regions associated with memory and learning #WCPG2022

ENIGMA GWAS in one month old infants shows variant association with brain volumes correlated with adult brain volumes. Correlation with SCZ not shown, but ongoing area of interest #WCPG2022

Similarly can find variant associations with longitudinal changes in brain structures and volumes #WCPG2022

Lots of working groups, biased to psych but also cancer etc. Tend to start with cortical and subcortical regions, then white matter volumes. #WCPG2022

Lots of disorder focussed work, but also baselining of expected volumes in healthy individuals. Case-control differences in regions seem to cross psych diagnostic boundaries - no ADHD specific regions etc. Pattern similar to genetic correlation between disorders #WCPG2022

Are observed differences due to shared genetic factors with disorders? Evidence is mixed - don't see this for SCZ, but have e.g. for ADHD. Ongoing work #WCPG2022

Built web database to examine genetic effects on brain structures, e.g. looking at rs4702 [FES/FURIN variant, in my opinion the most interesting variant in psych at the moment]. Can also generate brain plots (which increase publishability) #WCPG2022

ENIGMA has made the neuroimaging community more collaborative and increased co-publication (but UK Biobank has changed that a bit) #WCPG2022

Where is ENIGMA going? More detailed understanding of genetic influences on structure and function and role in psych - specific hypotheses to explore. #WCPG2022

Where is ENIGMA going? Treatment, both in terms of deep brain stimulation etc., but also blood brain barrier permeability #WCPG2022

Where is ENIGMA going? Diversity. Even harder to achieve than in psych genomics, because need to align and map to reference. Mostly based on MNI reference, based on 152 young adults. Issues with age. Issues with ethnicity - mostly EUR #WCPG2022.

Degree of image morphing to reference is heritable, and has real GWAS hits with relevant biology #WCPG2022