Modes of Brain Aging - Resources
...back to main FMRIB UK Biobank resource page
Below are online materials relating to our paper on multiple modes of
brain aging.
In all GWAS tables and summary statistics files, "AF" refers to the
frequency of A2 (which most often is the minor allele), and the (sign
of the) GWAS beta relates to the A2 allele count.
Results spreadsheets
These spreadsheets can be downloaded as Excel files or viewed
online as Google Sheets. Many of these have multiple sub-sheets,
with one tab (at the bottom of the view) per mode/mode-cluster. For
nIDP (non-imaging-derived phenotype) associations, -log10P, r and N
(samples used for correlations) are given, for all subjects, and
also for just females and just males; the VariableID refers to a
variable coding that can be looked up on UK Biobank
showcase.
- List of all IDPs (imaging-derived phenotypes): excel, online
- Associations of nIDPs with different groups of modes /
mode-clusters:
- IDP weights from modes / mode-clusters:
- GWAS results (peak associations) excel, online. Different
groups of results are arranged in several tabs at the bottom - see
below for the "V00...." phenotype naming explanation.
GWAS Manhattan plots
Each mode or mode-cluster results in a separate GWAS and hence a
separate Manhattan plot. Each PDF file of Manhattan plots shows the
results from one GWAS per page. The organisation of pages is:
- V0001-V0062 - brain-age delta estimates from 62 modes.
- V0063-V0124 - brain-age delta estimates from 62 modes (partialled delta).
- V0125-V0130 - brain-age delta estimates from 6 mode-clusters.
- V0131-V0136 - brain-age delta estimates from 6 mode-clusters (partialled delta).
- V0137 - brain-age delta estimates from all-in-one modelling using all
3,913 IDPs (uncorrected for age bias, 10-fold cross-validated).
- V0138 - brain-age delta estimates from all-in-one modelling using all
3,913 IDPs (corrected for age bias, 10-fold cross-validated).
- V0139 - brain-age delta estimates from all-in-one modelling using all
62 modes' deltas averaged.
- V0140 - brain-age delta estimates from all-in-one modelling using all
62 modes' deltas in a single GLM (corrected for age bias, 10-fold cross-validated).
- V0141 - sanity-check GWAS, of Left Putamen T2* IDP as shown in Elliott, Nature, 2018.
GWAS additional analyses
- Peak SNP lists For the full list of peak associations (SNPs), these results list genomic loci, the loci peak SNPs, and all independent SNPs (though note that because only peak associations were input to the FUMA analyses, these tables largely contain the same entries as each other).
- Annotated expanded SNP results These
results provide annotations for all SNPs in LD (linkage
disequilibrium) with the peak SNPs listed above.
- eQTL mapping
- Chromatin interactions
- PHEWAS For all of the peak
associations between modes/mode-clusters and SNPs, this lists other
traits previously reported (in the GWAS literature) as being
associated with those SNPs: tsv, online, help
- Genetic SNP-based (co)heritability via LD score regression
GWAS summary statistics
SNPs included here were preselected requiring that MAF>=0.001
(though results reported in the paper were more
stringent). Phenotypes 1-141 are as described above.
- b.txt betas (one row per SNP, one column per IDP, no header
row)
- s.txt error standard deviation (ditto)
- z.txt z statistics (ditto)
- p.txt -log10p (ditto)
- stats.txt other information, with headers on the first
row: chr rsid pos a1 a2 af info.
- See BGENIE
documentation for more information.
Resting-state functional MRI summary images
These maps are summaries of the mode and mode-cluster ICA
weights. "nodeamps25" and "nodeamps100" refer to the resting-state
node amplitudes from the low- and high-dimensional group-ICA
originally carried out on behalf of UK Biobank, explained and
illustrated in greater detail on
the UK Biobank brain
imaging resources page. Coloured numbers in the summary plots
show the ICA IDP weights (each node's amplitude is one
IDP). "edges25" and "edges100" show the functional connectivity edge
strengths, as explained and illustrated on the same resources
page. Each pair of images shows a node-pair, and the coloured bar
illustrates the strength and sign of the population-average
connectivity between these two nodes. The coloured numbers show the
ICA IDP weights (each edge is one IDP).
- Mode cluster 5 nodeamps25, nodeamps100
- Mode
17 nodeamps25, nodeamps100, edges100
- Mode 31 nodeamps25, nodeamps100
- Mode
33 nodeamps25, nodeamps100,
edges25, edges100
- Mode
41 nodeamps25, nodeamps100,
edges25, edges100
- Mode
45 nodeamps25, nodeamps100,
edges25, edges100
- Mode 52 nodeamps100
- Mode 58 nodeamps100, edges100
Code
Code from the Elliott Nature paper, for generating IDPs, nIDPs and
confounds, is here.
New code for all of the analyses in this paper is here.