Overview of the analysis procedure for this study

SuSiE (sum of single effects) is a univariate fine-mapping approach implemented within the R package susieR. ‘coloc’ is an R package for bivariate colocalisation analysis between pairs of traits. h 2 = Heritability, rg= bivariate genetic correlation. The analysis steps shaded in blue have been implemented within a readily applied analysis pipeline available on GitHub: https://github.com/ThomasPSpargo/COLOC-reporter.

Genome-wide association studies (GWAS) sampled

Each GWAS is a GWAS meta-analysis of disease risk across people of European ancestry. *Proxy cases from the UK Biobank cohort. †Estimated from cumulative risk after age 45 after correcting for competing risk of mortality and assuming a lifespan of ∼85 years. h2 = heritability

Comparison of genome-wide SNP significance against local genetic correlation significance thresholds in all trait pairs and loci analysed

All loci analysed showed sufficient local univariate heritability across compared traits to allow bivariate correlation analysi s. Subsequent fine-mapping and colocalisation analyses were performed in this study for regions with at least a false discovery rate (FDR) adjusted significance for the local genetic correlation. SNP = single nucleotide polymorphism.

Local genetic correlation analyses between trait pairs

The lower panel displays a heatmap of genetic correlations (rg) across genomic regions where any bivariate analyses were performed; white colouring indicates that the region was not analysed for a given trait pair owing to insufficient univariate heritability in one or both traits. The upper panel shows a Manhattan plot of p-values from each correlation analysis, denoting trait pairs by colour and comparisons passing defined significance thresholds by shape (square for a strict Bonferroni threshold and triangle for a false discovery rate (FDR) adjusted threshold); the hatched line indicates the threshold p-value above which Pfdr <0.05. The panels are both ordered by relative genomic position, with bars above and below indicating each chromosome. AD = Alzheimer’s disease, ALS = amyotrophic lateral sclerosis, FTD = frontotemporal dementia, PD = Parkinson’s disease, SZ = schizophrenia. Table S1 provides a complete summary of local genetic correlation analyses performed.

Colocalisation analysis conducted across 95% credible sets identified during univariate fine-mapping of trait pairs

N SNPs refers to the number of SNPs present for both traits and the 1000 genomes reference panel in the region within colocali sation and fine-mapping analysis. *Indicates comparisons with genetic correlation analysis p <8.26×10-5 (0.05/605). Δ Denotes locus extended by ±10kb for fine-mapping and colocalisation analysis. †Variant identified in colocalisation as having the highest posterior probability of being shared variant assuming hypothesis 4 is true (see Figure 3). §Differences in fine-mapping solutions across trait pairs in the Chr6:32.21-32.45Mb locus reflect differences in the SNPs retained after restricting to those in common between the compared GWAS ø H0 = no causal variant for either trait, H1 = variant causal for trait 1, H2 = variant causal for trait 2, H3 = distinct causal variants for each trait, H4 = a shared causal variant between traits. PIP = posterior inclusi on probability. AD = Alzheimer’s disease, ALS = amyotrophic lateral sclerosis, FTD = frontotemporal dementia, PD = Parkinson’s disease, SZ = schizophrenia.

Evidence for colocalisation between amyotrophic lateral sclerosis (ALS) and Alzheimer’s disease (AD) in the Chr6:32.63-32.68Mb region

Panel A: SNP-wise p-value distribution between ALS and AD across Chr6:32.63-32.68Mb, in which colocalisation analysis found 0.90 posterior probability of the shared variant hypothesis (see Table 3). Panel B: (upper) Per-SNP posterior probabilities for being a shared variant between ALS and AD, (lower) positions of HGNC gene symbols nearby to the 95% credible SNPs. Posterior probabilities for being a shared variant sum to 1 across all SNPs analysed and are predicated on the assumption that a shared variant exists; 95% credible SNPs are those spanned by the top 0.95 of posterior probabilities. The x-axis for Panel B is truncated by the base pair range of the credible SNPs and genomic positions are based on GRCh37.