Supplementary Information for Human Genetics
Pervasive pleiotropy between psychiatric disorders and
immune disorders revealed by integrative analysis of multiple
GWAS
Qian Wang1,2,3,† · Can Yang 3,4,5,† · Joel Gelernter 2,3,6,7, · Hongyu Zhao1,4,7,8,* †
†
These authors contributed equally to this work.
1 Program in Computational Biology and Bioinformatics, Yale University, New Haven,
Connecticut, USA
2 Department of Psychiatry, Yale School of Medicine, New Haven, Connecticut, USA
3 VA CT Healthcare Center, West Haven, Connecticut, USA
4 Department of Biostatistics, Yale School of Public Health, New Haven, Con- necticut, USA
5 Department of Mathematics, Hong Kong Baptist University, Hong Kong SAR
6 Department of Neurobiology, Yale School of Medicine, New Haven, Connecti- cut, USA
7 Department of Genetics, Yale School of Medicine, West Haven, Connecticut, USA
8 VA Cooperative Studies Program Coordinating Center, West Haven, Con- necticut, USA
† hongyu.zhao@yale.edu
S1 Fig
Number of publications. Number of publications from PubMed search of “Psychiatric disorders AND
autoimmune diseases” until Oct. 2014 per year.
S2 Fig
Conditional Q-Q plot showing the cross-enrichment of SCZ-MS GWAS signal but not BPD-MS
GWAS signal. Black dots represent all 287 757 SCZ or BPD GWAS SNPs overlapped with MS GWAS
SNPs. The other 4 colored dots represent different subsets of SNPs conditional on MS GWAS p-values
smaller than 0.05, 0.01, 0.001, 0.0001, respectively, with the number of SNPs of each subset shown in
brackets. Right panel shows Q-Q plot of SCZ GWAS conditional on MS GWAS; left panel shows Q-Q plot
of BPD GWAS conditional on MS GWAS. Top panel included all overlapped SNPs; bottom panel have
MHC region (25Mbp-34Mbp on Chromosome 6) excluded. It suggests that MS may have strong pleiotropy
with SCZ but not BPD, with or without the MHC region.
S3 Fig
Enrichment of CNS SNPs and immune eQTL with and without the MHC region. Red bars are
enrichment ratios estimated using all SNPs genome-wide, blue bars are enrichment ratios estimated with
SNPs in the MHC region (25Mbp-34Mbp on Chromosome 6) excluded.
S4 Fig
Enrichment of immune-eQTLs in SNPs shared between five psychiatric disorders and CD. Red bars
𝑞
represent the enrichment ratio of immune related eQTLs in 5 PGC traits specific SNPs respectively ( 10),
𝑞00
𝑞01
blue bars represent the enrichment ratio of immune related eQTLs in SNPs associated only with CD (
𝑞00
),
grey bars shows the enrichment ratio of immune related eQTLs in SNPs associated with both CD and the
𝑞
𝑞
corresponding PGC disorder ( 11). Among the immune related eQTLs, the ratios 11 are SCZ-CD 3.9 (s.e.
𝑞00
𝑞00
0.06), BPD-CD 4.4(s.e. 0.09), ASD-CD 3.4 (s.e. 0.2), MDD-CD 4.6(s.e. 0.18), and ADHD-CD 2.1(s.e.
𝑞
𝑞
𝑞
𝑞
0.45). The fact that 11> 10 and 11> 01, suggests the shared genetic components between the five
𝑞00
𝑞00
𝑞00
𝑞00
psychiatric disorders and CD are closely related to immune function.
S5 Fig
Enrichment of DNase-peak located SNPs in SCZ GWAS signal from 98 ENCODE cell lines. 98 cell
lines ordered by enrichment ratios; cell lines from blood and brain are colored red and blue respectively.
S6 Fig
Enrichment of H3K9ac in GWAS traits across different tissues. H3K9ac histone marker information
was collected from ROADMAP for eight tissues: blood, brain, breast, fat, heart, lung, muscle and skin.
Each bar represent the enrichment of a certain cell with the color indicating the tissue it belongs to. X-axis
indicates the 17 GWASs and y-axis shows the enrichment ratio. Both psychiatric disorders (top panel) and
immune-related disorders (middle panel) have the highest enrichment for H3K9ac markers in blood, while
educational traits (both years of education and colleges completion) have the highest enrichment for
H3K9ac markers in brain.
S7 Fig
Conditional Q-Q plot showing enrichment of SNPs having same effect direction between SCZ and
CD. Black dots for all 928 987 SNPs, and the other three lines are different subsets of SNPs, selected by:
blue for 472 165 SNPs that have same effect direction for SCZ and CD; red for 7 414 SNPs with p<0.001
in CD GWAS; green for the intersection of blue and red.
S8 Fig
Trend of consistent effect directions for shared SNPs of SCZ-RA and BPD-RA disease pairs. For each
disease pair, SNPs are assigned to 10 groups based on their posterior probability of being associated with
both diseases, and then the proportion of alleles having the same effect direction for the disease pair was
calculated within each of the ten SNP groups. Left, SCZ-RA; right, BPD-RA. Blue represents the same
effect direction, and red represents the opposite direction, x-axis represents proportion of SNPs.
S9 Fig
LD block grouped SNP effect direction, posterior probability and block size. For each trait pair, all
overlapped SNPs were grouped into 20 573 LD blocks. Each LD block was represented by a circle
positioned with x-axis being the maximum posterior probability of being associated with both traits, and yaxis being the proportion of SNPs having the same effect direction for two traits. The radiation of each
circle was scaled to be proportional to the square root of number of SNPs in a LD block. Top left: SCZheight; top right: SCZ-CD; bottom left: BPD-height; bottom right: BPD-CD. LD blocks with high
maximum posterior probability are more likely to have higher proportion of SNPs with same effect
direction for SCZ-CD and BPD-CD, while more random for SCZ-height and BPD-height.
S10 Fig
LocusZoom showing distribution of posterior probability in cytoband 1p13.2. Posterior probability of
being associated with both diseases in eight disease pairs between SCZ, BPD, MDD, ASD, and RA, T1D,
are shown separately. Only region 1p13.2 is shown. SNP with the highest posterior is labeled, and SNPs in
LD with it are colored. Figures plotted using LocusZoom.
S11 Fig
Protein-protein interaction enrichment. Constructed using the top 1000 SNPs, with color indicating
significance level.
S12 Fig
Posterior of SNPs being associated with both diseases for 28 disease pairs. Posterior of all shared SNPs
of a disease pair is plotted against 22 genomic positions. Red line indicates posterior=0.8.
S13 Fig
The frequencies of top SNPs (posterior probability > 0.8) appearing in 28 disease pairs. A total of 4
149 SNPs have posterior probability > 0.8 of being associated with at least one disease pair. Histogram
showing the number of disease pairs that these 4 149 SNPs are associated with (with posterior probability >
0.8).
S1 Table
Strength of pleiotropy within and across two disorder classes: 5 psychiatric disorders and 7 autoimmune
disorders.
S2 Table
A description of 98 ENCODE cell lines, and the enrichment ratio of DNase-peak located SNPs in SCZ
GWAS signals from each cell line.
S3 Table
Number of SNPs in effect direction analysis and proportions of SNPs in each direction combination
categories.
S4 Table
Genomic information and effect direction of 85 SNPs, whose posterior probability of being associated with
both SCZ and CD above 0.9.
S5 Table
Significance of cross-disorder cytoband enrichment tests.
S6 Table
Genome annotation enrichment results (adjusted p-value > 0.05), including gene ontology terms and KEGG
pathways.
S7 Table
A complete list of central nervous system (CNS) genes.