Corrected: Author correction ARTICLE https://doi.org/10.1038/s41467-019-09773-y OPEN A systems biology approach uncovers cell-specific gene regulatory effects of genetic associations in multiple sclerosis International Multiple Sclerosis Genetics Consortium 1234567890():,; Genome-wide association studies (GWAS) have identified more than 50,000 unique asso- ciations with common human traits. While this represents a substantial step forward, establishing the biology underlying these associations has proven extremely difficult. Even determining which cell types and which particular gene(s) are relevant continues to be a challenge. Here, we conduct a cell-specific pathway analysis of the latest GWAS in multiple sclerosis (MS), which had analyzed a total of 47,351 cases and 68,284 healthy controls and found more than 200 non-MHC genome-wide associations. Our analysis identifies pan immune cell as well as cell-specific susceptibility genes in T cells, B cells and monocytes. Finally, genotype-level data from 2,370 patients and 412 controls is used to compute intra- individual and cell-specific susceptibility pathways that offer a biological interpretation of the individual genetic risk to MS. This approach could be adopted in any other complex trait for which genome-wide data is available. A full list of authors and their affiliations appears at the end of the paper. Correspondence and requests for materials should be addressed to S.E.B. (email: Sergio.Baranzini@ucsf.edu) NATURE COMMUNICATIONS | (2019)10:2236 | https://doi.org/10.1038/s41467-019-09773-y | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09773-y T ranslating Genome-wide association studies (GWAS) dis- Roadmap projects (REP) to identify cell-specific networks likely coveries into functionally relevant biology has proven to be affected by the susceptibility variants (Fig. 1a, Supplementary highly challenging. The extensive linkage disequilibrium Fig. 1). We included all genome-wide (GW) significant single (LD) which typically flanks common variants means that most nucleotide polymorphisms (SNPs) together with their proxies GWAS identified SNPs are likely to be tags for functionally selected at differing LD thresholds (r2 > 0.5 was used for the main relevant variation rather than exerting any meaningful effects analysis). themselves. Furthermore, since the vast majority of associations All regulatory information retrieved for each MS-associated identified by GWAS map to non-coding regulatory regions it is region was compiled for each cell type in a single master likely that the underlying functionally relevant variants only exert table (Supplementary Data 1). This catalogue contains all of the pertinent effects on gene expression in particular tissues1–4. available regulatory features that potentially modulate the Fortunately, better powered studies, have increased the number of expression of each of the genes mapping to each associated associations identified enabling biological meaning to be inves- region in a cell-specific manner. We next used this information to tigated in aggregate (i.e. pathway analysis). In its simplest form, build a cell-specific, genetic regulatory network that constituted genes lying closest to the most strongly associated (lead) SNP the basis of a gene prioritization scheme within each associated identified for each association can be grouped into pathways or region (Fig. 1b). Specifically, each gene within a given locus specific functional memberships via the use of pre-assembled received a score (PRE) that does not depend solely on the closest controlled vocabularies (Gene ontology, KEGG, etc)5–8. This SNP, but that is equal to the weighted sum of all regulatory approach can be enhanced by using protein interaction networks features potentially affected by variation at nearby associated to more rigorously assess which of the candidate genes encode SNPs (See Methods). Figure 1c shows a heat map representation proteins that physically interact in any particular pathway9. Using of the PREs for all genes (n = 2,444) implicated by the GW this refined approach, we and others have been able to show that significant MS associations for each of the main cell/tissue types MS-associated genes are indeed more likely to interact in protein (Individual PREs for each cell/tissue and GWAS statistical space10–12. Furthermore, this analysis can be extended to include confidence are listed in Supplementary Data 2-28). Two GW association signals below the genome-wide threshold of sig- significant regions (10 and 21) are shown in larger detail as nificance and thereby nominate new additional potentially representative examples (Fig. 1c). Because it integrates actual meaningful associations11,13,14. However, the networks of genes/ regulatory information for each associated SNP and those in LD, proteins identified by these approaches are not cell- or tissue- this approach can prioritize the most likely genes affected by the specific, thus limiting the usefulness and interpretability of this same association signals in each cell type analyzed (Supplemen- information. tary Data 29). In the example, the PREs at the top of Fig. 1c With the completion of efforts like the Encyclopedia of DNA highlight the lead variant defining Region 10 (rs6670198, Chr 1) elements (ENCODE) and the Roadmap Epigenomics Project and those in LD which are likely to affect the expression of (REP) a wealth of information on regulatory elements is now FAM213B and TNFRSF14 in all immune cells (B, T and available from hundreds of cell types and dozens of different monocytes -M-). These SNPs would have almost no effect at all tissues15,16, raising the possibility of applying network-based in the CNS and lungs (L), as MS-associated variants are unlikely approaches in a cell-specific manner. We reasoned this approach to alter their expression in those tissues. For comparison, a simple would likely be highly informative in diseases like multiple proximity approach would have just implicated the FAM213B sclerosis (MS) where substantial numbers of associated variants gene, which, while being of biological interest, may describe an have been identified. MS is an autoimmune disease of the central incomplete scenario. Interestingly, TNFRSF14 (a member of the nervous system (CNS) and leads to a neurodegenerative process. TNF receptor superfamily) encodes a protein involved in signal Our recent GWAS meta-analysis and follow-up study have transduction pathways that activate both inflammatory and revealed a total of 233 genome-wide significant associations and a inhibitory T-cell immune responses due to its particular ability further 416 variants potentially associated with MS17. to interact with multiple ligands in distinct configurations18. Here we develop a framework to interpret such associations in Another example is provided by region 21, defined by the lead the context of cell-specific protein networks to identify the most SNP rs6032662 mapping to chromosome 20. rs6032662 maps likely process(es) affected by the non-MHC associations as a midway between NCOA5, a tumor suppressor gene, and CD40, whole. This approach involves (1) selecting independently asso- which encodes a well-known member of the TNF-receptor ciated signals in extended haplotypic blocks; (2) identifying the superfamily. This receptor is essential in mediating a broad genomic regulatory processes likely to be altered by the poly- variety of immune and inflammatory responses including T cell morphisms in these blocks in a cell-specific manner; (3) com- activation, T cell-dependent immunoglobulin class switching, puting a cell-specific gene score for genes in each associated locus; memory B cell development, and germinal center formation19. (4) building cell-specific gene/protein networks; (5) Interpreting While its expression is highest among antigen presenting cells the biological processes most likely affected for each of the cell (including those derived from dendritic cells and monocyte/ types studied. We demonstrate this approach in the latest GWAS macrophages) its PRE score is low in the M set (Fig. 1c), meta-analysis in MS involving a total of 47,351 cases and 68,284 indicating that, of the cells studied, MS-associated variants only controls. Furthermore, we use genotype-level data from a subset regulate its expression in B cells. Altogether, these results show of 2370 cases and 412 controls to identify cell-specific intra- that this approach is a useful strategy to prioritize genes within individual risk pathways. These individualized scores can be used association regions. as a global risk measure in subsequent associations with more detailed phenotypes. Protein connectivity among products of MS-associated loci. Previous studies have shown that the proteins encoded by genes Results affected by genetic association signals are more likely to interact, Predicted regulatory effects (PRE) of MS-associated variants. in part because they often participate in the same biological We integrated genetic association signals from the latest genetic pathways6,20–22. Thus, we evaluated (for each cell/tissue) how analysis of MS17 with cell specific information on regulatory many of the gene products in MS-risk loci predicted to be posi- elements available from the ENCODE and Epigenomics tively regulated (i.e. PRE > 25th percentile or PRE-25) also 2 NATURE COMMUNICATIONS | (2019)10:2236 | https://doi.org/10.1038/s41467-019-09773-y | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09773-y ARTICLE a GWAS (summary-level data) PPI Cell-specific population-level genetic networks Cell-specific gene PRE regulatory info Intra-individual GWAS genetic networks (individual genotypes) b Regulatory marks Regulome DBTM 0.5 0.5 report for GWAS SNPs SNP # 123 Transcriptional activator Gene 1 Enhancer Cell type Reg feature Target gene Repressor B cell H3K4Me3 Gene 1 4 SNP 1 B cell H3K27Ac Gene 2 B cell TFBS Gene 2 T cell DHS Gene 1 2 Monocyte H3R2M Gene 3 2 1 Gene 2 Activators Inhibitors Net weight PRE Gene 1 4 1 3 7 0.5 Gene 2 3 0 3 3.5 c Region 21 Region 10 p1 3 21 q1 12 20 .13 33 p1 2.3 20 .1 20 1.2 2 .1 1. 11 21 20 23 20 13. 3. 3. 3 2 2 3 1 3 2 q1 2 12 41 q4 3 31 p1 p1 p1 p1 q1 q1 q1 2. 1. 2. 1. .2 1. 4 4 q 33 1q q3 1q 1q p3 1p 12 20 20 20 q1 20 20 chr20 chr1 (q13.12) (p36.32) PRE Low High Region 21 (rs6032662) Region 10 (rs6670198) 50 kb hg38 200 kb hg38 ACTRT2 CD40 FAM213B LOC115110 SKI MORN1 TNFRSF14 MMEL1 ACTRT2 MMP9 RP5-998H6.2 MANBA RP13-436F16.1 RP11-740P5.2 PLCH2 CD40 NCOA5 MELL1 TTC34 RP3-395M20.7 MMEL1 NCOA5 RP4-740C4.7 HES5 NFKB1 MMEL1 TNFRSF14 RP3-395M20.3 RP11-465L10.10 SLC12A5 TTC34 RER1 PANK4 SLC12A5 UBE2D3 FAM213B RPL13P2 PEX10 RP3-395M20.8 B T M C L B T M C L B T M C L FAM213B Fig. 1 Overall strategy and computation of the predicted regulatory effect (PRE) in MS-associated loci. a GWAS signals were integrated with cell-specific regulatory information to compute PRE at both population and individual level. In a second stage, genes with high PRE at each of the cell types analyzed were identified in a human protein interactome (PPI) and sub-networks of enriched genes (proteins) were extracted. b Each MS-associated SNP and those in LD were used as query in RegulomeDB. For each SNP, the all regulatory features were annotated and classified according to type and cell of origin. A graph connecting every queried SNP (crosses), the regulatory feature (diamonds), and the target gene (circles) was created and the number of experiments supporting a particular regulatory feature was used as weight (numbers next to SNP). Finally, a PRE score was computed for each gene by summing up weights from all incoming regulatory signals for each of the cell types analyzed. c Heatmap represents the PRE of all genes under GW MS- associated loci for cells of interest. Rows represent genes, and columns denote cell types. Colors indicate positive (red), neutral (white) and negative (blue) PRE values. Two representative regions are highlighted. Region 10 (associated SNP: rs6670198, green box) highlights immune-specific (B, T, and M) regulation of FAM213B and TNFRSF14. In contrast, region 21 (associated SNP rs6032662, blue box), shows high PRE only for CD40 in B cells. C: CNS; L: lung; T: T cells; M: monocytes; B: B cells. This analysis represents all SNPs with an r2 > 0.5 of the main GW effect interacted in a human protein network containing 15,783 pro- The three significant networks (Fig. 3a–c) shared several of their teins and 455,321 interactions (See methods) (Fig. 2 shows a genes, thus constituting a core module (Fig. 3d). The molecular schematic of this approach). In addition to the total number of functions of most of these genes belonged to the binding (36%) interactions we also computed other relevant network metrics and catalytic activity (33%) categories. Other functions were such as the size of the largest connected component –LCC– and receptor (13%), signal transduction (6%), and structural molecule the number of connections –edges- among nodes within the LCC. (10%). A PANTHER analysis revealed these genes belong to JAK/ These metrics, if statistically significant, are often good indicators STAT, IFNgamma, interleukin, and integrin signaling pathways, of true biological networks. We found that, for all immune cell among others. Altogether, our analyses suggest that susceptibility types analyzed, the above network metrics among GW associated to MS stems from a core of processes that can be active in any of gene products always exceeded those from 10,000 randomly several immune-related cell types. These findings are in agree- generated, size-matched networks. Specifically, network metrics ment with the “omnigenic” model of inheritance of complex among the gene products of GW associated loci were statistically diseases, posing that gene regulatory networks are sufficiently significant for T cells, B cells and monocytes (Fig. 3). The number interconnected such that all genes expressed in disease-relevant of interactions among genes related to the CNS were not sig- cells are liable to affect the functions of core disease-related nificantly different than expected. One factor likely affecting this genes23. result is that in contrast to immune cells (for which PREs were As predicted by the omnigenic model, we noted that several computed on a cell-specific manner) computations for CNS are genes were only present in some cell types but not others. For the result of 25 different cell types/anatomical regions. This could example, CD28 was only present in the T cell network, ELMO1 in potentially smooth the overall estimate of PREs as the various cell B cells and MERTK in the monocyte/macrophage lineage. CD28 types within the CNS could be under different regulatory control. is located on the surface of T cells and provides a required NATURE COMMUNICATIONS | (2019)10:2236 | https://doi.org/10.1038/s41467-019-09773-y | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09773-y Genes with high T B regulatory potential M p < 10–5 C PPI 15,783 nodes 455,321 edges Extract cell-specific sub-network 109 nodes 0 50 71 71 edges # edges Fig. 2 Network connectivity analysis. The PRE of genes were loaded as attributes in a protein interactome. In the central panel, genes with a PRE above the 95th percentile of their respective cell-specific distributions are visualized (M: monocyte, green; T: T cells, red; B: B cells, blue, C: CNS, yellow). For each cell type, the number of edges in the sub-network composed of interacting proteins with PRE above the threshold was analyzed. In this example, the CNS sub- network is composed of 109 nodes and 71 edges. Ten thousand random networks with the same number of nodes (i.e. 109) were generated and the distribution of edges was plotted along with the number of edges of the relevant sub-network (i.e. 71). A p-value was computed to evaluate the probability that this number of edges was seen by chance a B cells b T cells IDE 15/50 (30%) B cell exclusive 12/54 (22%) T cell exclusive Membrane Membrane CD48 IL2RA S1PR1 * CD69 SLAMF7 * * * CLEC2D IL7R RPS6KB1 IQGAP1 TAGAP SLC6A16 CSF2RB SLAMF1 CD28 CTLA4 STAT3 CD5 8 TAGAP RHGAP31 RGS1 MADD CD69 * IL7R * CXCR4 SLAMF1 CD58 IDE TXK S1PR1 * LPP CD86 CD48 CLEC2D * * MPHOSPH9 CD226 IL2RA TGFBR3 ELMO1 * * PLEK * TNFRSF1A * GRB2 PLEC RPS6KB1 VMP1 ZFP36L1 PLEK GRAP2 * CAMK2G ACTN1 ZFP36L1 SART1 200 0.00209 PLEC MAPK1 GRB2 PPM1F SART1 ACTN1 GRAP2 200 0.00884 NDFIP1 * NANS 150 CAMK2G 150 Frequency MAPK1 Cytosol Frequency PPM1F Cytosol GOLGB1 * CCNE2 * * 100 100 * * * STK38 DAB2 KIF11 PAPD7 MYNN L PP HHEX * MYB 50 MYB ZFP36L2 STAT4 RUNX3 50 ZMIZ1 IQGAP1 FOXP1 SMARCA4 SP140 PAPD7 BATF CBLB STAT3 DDX6 SP140 Nucleus * SMARCA4 JAK1 RCOR1 ZFP36L2 TEC DDX6 0 Nucleus FOXP1 DAB2 TCF7 MYNN STK38 JAK1 CBLB ZMIZ1 * TRIM14 0 0 50 100 150 * * # of edges * 0 20 40 60 # of edges 80 100 120 c Monocytes d Common module 3/48 (6%) Monocyte exclusive CD48 Membrane CD58 Membrane S1PR1 S1PR1 CD48 TAGAP TAGAP CD58 CLEC2D IL7R CD69 CLEC2D IL7R CD69 CSF2RB TNFRSF1A MADD RGS1 MPHOSPH9 PLEK PLEK RPS6KB1 ELMO1 PLEC ZFP36L1 MAPK1 * NCF4 RPS6KB1 IDE 0.00109 Binding (GO:0005488) Catalytic activity (GO:0003824) GRB2 PLEC IDE GRB2 ACTN1 600 Receptor activity (GO:0004872) MAPK1 PPM1F CAMK2G CAMK2G SART1 CCNE2 MERTK Signal transducer activity (GO:0004871) SART1 PPM1F ACTN1 Cytosol * Structural molecule activity (GO:0005198) Cytosol Frequency 400 LPP TEC IQGAP1 ZFP36L1 ZFP36L2 200 LPP IQGAP1 MYB DAB2 ZFP36L2 STK38 SMARCA4 DAB2 STK38 MYB SMARCA4 KIF11 JAK1 BATF PAPD7 JAK1 ZMIZ1 PAPD7 0 STAT3 MYNN ZMIZ1 Nucleus TXK STAT3 MYNN RUNX3 TRIM14 CBLB DDX6 CBLB DDX6 Nucleus FOXP1 0 20 40 60 80 FOXP1 # of edges Fig. 3 Cell-specific gene sub-networks of GW associated regions (r2 > 0.5). Graphs correspond to the largest connected component in each cell/tissue bucket. Nodes represent proteins and edges represent interactions. For each cell type the PRE is proportional to the color intensity (dark: high; light: low). Genes/proteins are organized according to their cellular distribution. The histogram next to each sub-network shows the distribution of the number of edges of 10,000 randomly generated networks. The red arrows denote the number of edges observed in the corresponding sub-network and the p-value, the probability of observing a more extreme number of edges in a randomly generated network. a B cells; b T cells; c monocytes. An asterisk is placed next to genes/proteins exclusively observed in that cell type. d shows an aggregate (common) module present in all three cell types. A pie chart describes the GO: molecular functions assigned to these genes and a table describes the nine PANTHER pathways that were significantly enriched co-stimulatory signal to trigger their activation after engagement that transduces signals from the extracellular matrix into the of the MHC-antigen-T cell receptor trimolecular complex24. cytoplasm regulating many physiological processes including cell ELMO1 encodes a cytoplasmic adaptor protein that interacts with survival, migration, differentiation, and phagocytosis of apoptotic DOCK family guanine nucleotide exchange factors to promote cells (efferocytosis). Specifically, MERTK plays several important activation of the small GTPase RAC, thus enabling lymphocyte roles in normal macrophage physiology, including regulation of migration25. The MERTK gene encodes a receptor tyrosine kinase cytokine secretion and clearance of apoptotic cells26. These 4 NATURE COMMUNICATIONS | (2019)10:2236 | https://doi.org/10.1038/s41467-019-09773-y | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09773-y ARTICLE observations suggest that, in addition to the core susceptibility personalized approach takes into account the specific risk alleles module, at least part of the risk is cell type-specific. present in each individual and thus enables exploration of subject We next performed a sensitivity analysis by testing separately heterogeneity in a biological context and in a cell specific manner. GW, statistically replicated (SR), and non-replicated (NR) effects Hierarchical clustering of subjects and heatmap visualizations of (see Methods) SNP sets at three different LD cut-offs (r2 > 0.1, r2 the PRE of genes under regions 9, 49, and 53 (r2 > 0.5) for all cell > 0.5, and r2 > 0.8) and three PRE thresholds (PRE-10, PRE-25 types in this subset of cases and controls are shown as an example and PRE-50) (Supplementary Data 30). As anticipated, although (Fig. 4). The heterogeneity across individuals in the PRE of a the most significant network metrics were seen in the analysis given gene can be readily seen for all cell types. As expected for based on GW signals (Supplementary Fig. 2, panel A), some common variants, cases and controls (denoted by red and green significant network metrics were also seen in the analysis based horizontal bars in the leftmost column) do not cluster separately on the SR SNPs (Supplementary Fig. 2, panel B). This confirms within any single association region. This analysis provides a that in well powered studies variants with evidence for association visual representation of which genes are most likely affected by just short of GW significance may still represent real effects14. In common variants associated with the disease in those individuals, contrast, the connectivity of networks obtained with SNPs from in a cell-specific manner. The PRE for most genes in each the NR set was usually not significantly higher than that of heatmap show ample variability across individuals, highlighting random networks (Supplementary Fig. 2, panel C). As expected, genetic differences in their susceptibility to MS at this locus. including the wider range of SNPs implicated by relaxing the LD These maps also reveal which genes are most likely to be affected threshold down to r2 > 0.1 resulted in less significant network in each cell type by these common alleles. For example, while metrics, suggesting that including less robust proxies introduced associated variants near the gene EOMES (region 9) potentially more noise than signal. modulate its expression in T cells, this gene is less regulated in B cells and monocytes and strongly silenced in CNS (Fig. 4a). This is consistent with its function as a critical transcription factor in T Individualized PRE correlate with gene expression. The detailed cell differentiation27,28. Interestingly, higher PRE of these variants mapping of regulatory information for each SNP suggests that are observed for Th2 cells than for any of the other subsets if PRE are computed for a given cell type in a single individual analyzed. Previous reports revealed that EOMES expression limits based on the carriage of relevant risk alleles, these values FOXP3 induction, thus effectively reducing Treg populations29. should capture a non-negligible proportion of the variance in Genomic variation at this locus resulting in dysregulated EOMES gene expression in that cell type. To test this hypothesis, we expression in T cells and NK cells might be a critical mediator of interrogated the expression of the entire transcriptome of FACS- the risk to MS30. sorted CD4+ T cells, and CD14+ monocytes from 25 MS patients Another interesting example is the high PRE observed for the by RNAseq and then assessed the correlation of their genotype gene CD40 (region 21) preferentially in B cells for most subjects dependent PRE and their actual gene expression in each cell (Fig. 4b, pink boxes). This finding is consistent with the critical type separately. Our results showed that the correlation observed role of CD40 in B cell development and maturation, and was in all cases significantly higher than what would be expected indicates that MS risk affecting B cell biology is higher in some by chance if these metrics were independent. Furthermore, the subjects than others. Furthermore, individuals carrying the risk computed correlations were always higher for the matching variants within CD40 (rs4810485*T) have been reported to cell type (CD4/CD8 expression with T cells PRE and CD14 express lower levels of CD40 in the surface of their B cells and expression with monocytes PRE) (Table 1). The average corre- lower IL-10 levels31. This could carry therapeutic implications lation between RNA expression and PRE within the same cell considering the prominent role of B cell depletion therapy in type was 0.331 (CD4 vs. T cells, p < 10−300, linear regression), this disease32,33. Another signal revealing B cell involvement in 0.324 (CD8 vs. T cells, p < 10−300, linear regression), and 0.246 MS risk is the high PRE of all members of the Fc receptor-like (CD14 vs. monocytes, p < 10−300, linear regression), representing (FCRL) gene family in region 53 (Fig. 4c, yellow boxes) a significantly higher than expected value for each cell type. preferentially in B cells for most subjects, consistent with the Correlations between PRE and RNA expression of mismatched function of these gene family as regulators of proliferation in B cell types were significantly lower. These results suggest that the cells and phagocytosis34. computation of PRE can be applied to single patients and indi- vidual scores can be generated for each of them. We then used the genotype-level data from one of the GWAS Intra-individual risk networks are more connected in MS. datasets (UCSF, Supplementary Data 31), composed of 2370 Finally, we integrated individual risk regulatory scores with a patients and 412 controls, to compute cell-specific risk scores for global protein interactome to compute intra individual, cell spe- each individual using the same pipeline used for the population- cific risk networks. We hypothesize that these networks would level data. Rather than considering all 200 associations, this provide a step forward in the description of aggregate persona- lized risk scores by representing risk in a biologically relevant manner35,36. Furthermore, building risk profiles with pathway and cell specific information describes more accurately the biol- Table 1 Correlation between regulatory potential and cell ogy potentially affected by risk variants inherited by a particular specific global gene expression individual. We also hypothesized that similarly to what we observed for cases and controls at the population level, more PRE CD4 r (p-value) CD8 r (p-value) CD14 r (p-value) interactions among proteins encoded by risk loci would be Monocyte 0.225 (p < 10−300) 0.218 (p < 10−314) 0.246 (p < 10−300) observed for cases than for controls. This was indeed the case, as T-cell 0.331 (p < 10−300) 0.324 (p < 10−300) 0.253 (p < 10−300) we observed statistically significant differences between cases and B-cell 0.204 (p < 10−248) 0.204 (p < 10−247) 0.219 (p < 10−287) controls for the three main cell types tested (the CNS was not CNS 0.108 (p < 10−75) 0.112 (p < 10−82) 0.120 (p < 10−93) Lung 0.117 (p < 10−86) 0.122 (p < 10−94) 0.133 (p < 10−112) significant as shown in Fig. 5a and Supplementary Fig. 3) (Sup- plementary Data 32). The largest number of intra-individual Bold values indicate the corresponding cell type for the RNAseq results (e.g. CD14 RNAseq data interactions among gene products with high PRE was observed in corresponds to monocytes, and CD4 and CD8 corresponds to T cells) monocytes, followed by T cells, B cells and the CNS (Fig. 5a NATURE COMMUNICATIONS | (2019)10:2236 | https://doi.org/10.1038/s41467-019-09773-y | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09773-y a Region 9 b Region 21 c Region 53 B cells Monocytes B cells Monocytes B cells Monocytes PRE High FCRL4 FCRL5 FCRL1 FCRL2 FCRL3 FCRL4 FCRL5 FCRL3 FCRL2 FCRL1 Eomes Eomes CD40 CD40 CNS T cells Low CNS T cells CNS T cells FCRL5 FCRL3 FCRL4 FCRL2 FCRL1 FCRL4 FCRL5 FCRL3 FCRL1 FCRL2 Eomes Eomes CD40 CD40 Th1 Th17 Th1 Th17 Th1 Th17 Eomes Eomes FCRL4 FCRL3 FCRL5 CD40 CD40 FCRL4 FCRL3 FCRL5 Th2 Treg Th2 Treg Th2 Treg FCRL5 FCRL5 FCRL4 FCRL2 FCRL3 Eomes Eomes CD40 CD40 Fig. 4 Individualized PRE computations for three representative associated regions. Each row represents an individual (out of 2370 cases and 412 controls), and each column represents a gene within the associated region. Region 9 (a) contains the gene EOMES (green boxes), region 21 includes CD40 (pink boxes) (b) and region 53 (c) the FC receptor-like cluster (yellow boxes). The leftmost column denotes subject status (red: cases; green: controls) shows results for PRE-25). This is consistent with the larger percentile connectivity of T cell (51st), B cell (29th) and M (75th) significance of these risk networks observed at the population risk networks for this subject rank noticeably lower. level (Fig. 3). Interestingly, PRE values correlate with the global While the number of interactions (edges) among proteins polygenic risk (Supplementary Fig. 4) but it uniquely enables encoded by genes in associated loci was variable across cell identification of high-risk and low-risk individuals in a cell- types, on average, more interactions were observed for cases than specific manner. Figure 5b highlights four case:control pairs with for controls in all cell types studied. Supplementary Fig. 3 shows different risk profiles in each of the four main cell types/tissues the significance of testing different network parameters between analyzed. For example, subject 201100870 (case) is at the 99th cases and controls across a wide range of conditions. Similar to percentile of the distribution of network edges for monocytes what we observed at the population level, significant effects (120 nodes and 271 edges). In contrast, subject 20020214 (con- were also seen for loci with less than genome-wide evidence for trol) is at the 1st percentile (61 nodes and 98 edges). association, suggesting that some of those variants considered Another interesting observation emerging from this analysis is less strongly significant, also confer risk. that subjects at the extremes of the distribution of intra-individual In summary, this analysis underscores the notion that total MS interactions (a proxy for their overall risk) can be identified for risk is not only carried by accumulation of risk alleles, but also by each cell type. For example, subject 201327986 has 110 nodes and how the genes and proteins affected by those polymorphisms 190 edges in this subject’s B cell risk network (blue box), interact within each cell type. We anticipate this model could corresponding to the 99th percentile of all cases (Fig. 6). In apply to other common diseases. contrast, the corresponding percentile of the number of edges in his T, M and C (red/green/yellow) networks is substantially lower (47th/49th/66th). In line with results observed at the population Discussion level, individual CNS risk networks are consistently smaller and In this work, we provide evidence that integration of associated less connected than those from B cells, T cells and monocytes. variants from GWAS with regulatory information and protein Although CNS is still the least connected network in subject interactions, provide plausible models of disease pathogenesis. 201101897, with 118 edges in its CNS risk network (yellow box), This approach not only offers a data-driven solution to prioritize it ranks in the 99th percentile of all cases. In contrast the which genes within a locus are most likely affected by the risk 6 NATURE COMMUNICATIONS | (2019)10:2236 | https://doi.org/10.1038/s41467-019-09773-y | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09773-y ARTICLE a E-LCC b Monocyte B cell 2011009 28 2006146 83 2011021 52 2014110 33 2006140 99 2015010 55 2011025 46 2014204 73 2014107 67 2011024 19 2011018 82 2011022 49 2014106 56 2014107 12 2013278 88 2014205 13 2013279 61 2011024 90 39 2007242 20020094 2011022 95 2011021 37 2014108 83 2011026 41 2014107 24 2011019 43 2014110 06 2013278 78 2011021 54 2011024 92 2006142 59 2011020 83 2011021 91 2006146 47 2014110 04 2013280 93 20026683 2011010 76 2011009 63 2013279 56 2011024 18 2011006 85 2011009 42 2011024 24 2011026 75 2014111 33 2011019 65 2011011 03 2014111 18 2011021 19 2014110 03 2013281 49 20020243 2011025 63 2014205 18 2006142 67 2014108 66 2011021 17 2011010 89 2006147 52 2013279 58 2011021 51 2014110 54 2011022 43 2014109 36 2014204 77 2015010 56 2011020 22 2011020 35 2011008 45 2011010 81 2011022 14 2011024 96 2014107 94 2011026 44 2011014 68 2014106 21 2011026 14 2014108 44 2011024 95 2011021 65 2011009 57 2014109 37 2014106 57 2011009 23 2011009 01 2013280 29 2011025 62 2014108 56 2006155 66 2013281 44 2014204 29 20020272 2011026 19 2013280 23 2014204 84 2014111 28 2014111 69 2013280 54 2011011 52 2003085 08 2014111 74 2011025 31 2011007 85 2011012 06 2014107 87 2006147 26 2011023 97 2006142 88 2011023 93 2014204 56 19890147 2011026 87 2014107 50 2002112 96 2014204 44 2013279 92 2011010 18 2013281 47 2014106 82 2011010 05 2014108 71 2014111 10 19994445 2011024 09 2006146 09 2011008 80 2011015 02 2014205 14 2006141 53 2011024 69 2014110 20 2014106 97 2011008 72 2011020 80 2013279 03 2011010 96 2011019 78 2011016 47 2011021 24 2011010 04 2014109 35 2006145 98 2011018 59 2006147 75 2014107 68 2014111 32 2011025 82 2007237 37 2014111 44 2014107 07 2013279 04 2011022 97 2014204 26 2014110 77 2011014 10 2006141 64 2007240 12 2014109 50 19982776 2011021 35 2011019 15 2011019 37 2004057 41 2011018 57 2011012 05 2011019 57 2011006 84 2011021 56 2011006 40 2013280 78 2011022 56 2011020 57 2014108 27 2014107 76 2011019 26 2013278 96 2014110 95 2011018 69 2011022 29 2006142 72 2011019 09 2011016 64 2011023 05 2011006 71 2011026 98 2011008 75 2006147 84 2011024 62 2014108 48 2011022 23 2011009 59 2011023 01 2011012 01 2013278 32 2006138 52 2011025 54 2011022 31 2011021 81 2005045 55 2011025 06 2014109 27 2014111 66 2014107 96 2013281 24 2014204 68 2011027 02 2006140 76 2014108 47 2014106 87 2011027 46 2006141 80 2014109 13 2014108 70 2011020 43 2013278 81 2011018 47 2014109 52 2014109 68 2013278 69 2014111 14 2011019 63 2013279 55 2011010 23 2011009 11 2006142 95 2011016 99 2007233 27 2013280 82 2006155 62 2014106 88 2014108 01 2014109 81 2007237 35 2011020 76 2011025 44 2011021 77 69 2011025 2006148 33 2011010 83 2014110 47 2006141 30 2014107 29 2013278 68 2011024 53 2015010 92 2011022 86 2014110 50 2014108 86 2006146 20 2007236 53 2007233 47 2011024 77 2011026 88 2006145 86 2006145 70 2011008 11 2006155 71 N : 120 N : 110 2013279 26 2011025 75 2011021 00 2011025 66 2006138 00 2013281 67 2011020 01 2006148 36 2014110 18 2006147 31 2013281 39 2006145 75 2011007 30 2013279 05 2011006 99 2013281 66 2006147 99 2014110 38 2011020 28 2011020 47 2013278 40 2011025 67 2006144 77 2014205 07 2014107 66 2007233 97 2014108 09 2014107 39 2011026 28 2011009 71 2011013 03 2011018 77 2011023 10 2011026 03 2013280 13 2011014 36 2006145 81 2014107 90 2014106 69 2006142 23 19995542 2011021 32 2011026 39 2013280 33 2007237 09 2013279 48 2007239 78 2011014 25 2007237 08 2011027 01 2014108 90 2014107 58 2013279 06 20026678 2011015 97 2011013 25 E: 271 E: 190 2007236 56 2011019 55 2011018 27 2011023 08 2011019 01 2011023 09 2011027 36 2014110 09 2013279 68 2014110 31 19994491 CASE CASE 2013280 95 19881594 2013278 79 2014106 45 2014108 26 2011025 68 2011026 08 2011007 06 2013279 13 2011019 85 2006141 89 2014204 34 2011027 30 2014110 17 2006146 10 73 2011026 2011022 93 2004056 96 2011021 39 2011024 43 2011024 33 2013280 38 2011025 05 2011007 22 2011006 44 2003085 06 2006147 12 2011025 79 2014204 35 2013279 47 2011020 53 2011019 11 2011020 10 2011006 62 2014204 72 2011008 22 2014110 94 2013281 02 2011007 33 2013280 80 2007237 13 2014107 09 2004058 61 2011025 03 2011023 14 2014106 76 19891855 2011007 95 2011006 41 2014106 25 2013281 50 2015010 90 2013279 02 2014110 71 2011017 91 2014107 04 2014107 40 2011021 34 2013280 69 2014204 41 P : 99 P : 99 2014111 25 2011009 92 2011023 89 2011020 97 2014107 45 20020186 2011018 80 2014205 19 2014205 32 19994365 2011008 71 2006146 07 2013279 19 2011007 60 2014108 04 2014107 64 20011758 2014110 72 2011008 40 2011023 56 2011026 71 2014205 40 2011025 27 2014204 55 2014109 07 19960018 2011023 38 2013280 43 2014204 92 2014204 71 2014205 03 2011022 90 2011016 66 2014107 71 2014108 64 2011024 64 2007237 20 2006145 89 2006146 24 2006148 38 2013279 97 2011018 63 2014107 95 2013279 20 2011006 77 2011009 50 2011024 80 2011022 81 2011023 49 2011020 61 2013280 11 2011007 57 2013279 44 2011023 25 19975451 2011021 33 2011027 35 2014108 69 19994471 19891044 2014106 86 2011011 15 2011009 29 2011008 47 2011019 96 2014109 43 2014204 78 2013281 56 2011019 99 2011027 12 2011006 48 2011009 15 2011023 66 2011023 52 2011023 07 2011019 52 2006145 29 2011018 66 2014204 67 19994409 20026685 2014204 65 2014204 18 2011026 66 2014111 05 2013280 98 2013278 89 2011022 58 2011006 92 2007234 45 2011019 45 19994653 2011021 01 2014110 62 2013280 07 2011011 35 2013280 81 20020058 2011022 04 2011022 77 2011022 71 2006145 02 2013280 74 2006140 50 2011007 12 2011020 06 2014109 24 2011011 63 2011026 85 2006155 92 2011009 75 2014109 76 2011021 90 2014204 79 2011026 10 2014109 12 2011019 86 2014107 78 2007237 26 2007232 89 2011013 66 2011022 44 2011023 95 2014106 23 2011025 10 2013279 35 2014110 48 2014107 88 2014109 08 2014107 84 2014110 27 2014109 92 2006142 44 2011020 62 2011024 70 2014111 40 2011023 35 2011024 12 2011019 46 2011009 30 2014109 64 2014204 21 2006146 65 2014111 46 2011009 54 2011021 04 2014110 10 2011027 31 2014106 24 2006148 55 2011023 92 2013281 45 2011026 04 2011020 07 2011025 12 2014205 44 2011020 40 2014204 87 2003085 07 2014111 76 2011012 00 2013281 28 2011010 07 2014109 05 2011022 47 2014204 19 2011020 08 2011013 68 2011017 85 2014107 77 2013279 09 2013279 78 20020268 2013281 20 2006151 63 2011021 96 2011024 11 2014205 42 2014106 11 2014109 47 2007237 10 2014106 99 2013280 76 2014204 57 2013281 26 2011025 86 2011007 51 2011011 07 2006141 78 2011024 52 2011019 12 2014204 23 2014204 37 2011023 85 2013278 74 20020382 2014106 84 2013280 87 2014111 31 2011018 50 2011008 99 2004007 52 2014108 82 2011016 69 2011019 95 2014204 49 2011013 30 2014205 17 2004055 55 2011008 87 2006142 11 2014110 67 2013278 37 2006146 18 2011022 99 2014106 15 2014108 05 2011020 79 2011010 67 2014111 67 2011018 89 2005045 31 2004057 17 2014111 72 2013280 42 2011010 06 2011025 40 2011021 75 2011009 69 2014111 24 2007237 02 2006150 04 2011023 22 2014109 78 2014205 39 2011016 33 2014111 21 2011014 81 2014106 31 2011026 67 2011010 73 2011021 08 2014204 48 2011009 10 2006141 40 2011025 24 2011020 32 2013280 64 2015010 91 2011020 95 2014111 64 2014109 86 19982728 2011014 71 2011019 58 2014106 35 2011008 04 2011007 39 2015010 78 2014204 28 2006155 33 2011010 82 2004025 20 2011009 73 2006142 85 04 2014109 2014109 98 2011016 65 2011008 48 2011024 98 2011023 27 2011024 02 2014106 71 2013279 76 2013280 20 2011026 63 2014106 72 2011024 07 2011013 76 2014111 26 20020031 2011027 32 2011023 46 2011006 89 2014106 77 2013280 56 2011011 51 2011017 78 2011025 84 2014107 25 2011022 13 Subject_id: 201100870 2011022 62 2011014 53 2014205 01 2006141 13 2011024 21 2011007 75 2011020 41 2011007 80 2004058 16 2014111 06 2011024 06 2011018 68 2011023 33 2011013 55 2013279 12 2013278 75 2011024 67 2011008 60 2014204 66 2014109 58 2013280 97 2011023 47 2011021 66 2011025 07 2015010 61 2011008 37 2007232 19 2014107 43 2011017 83 2011022 73 2014204 76 2011021 44 2014204 32 2014111 08 2014109 87 2011006 91 2014107 01 2011018 24 2013281 09 2011010 43 2014110 92 2011026 16 2011026 07 19994307 2014204 86 2014108 73 2011007 82 2003085 01 2013278 86 2011021 50 2006142 54 2006147 88 2006155 07 2011020 31 20013644 2014106 60 2014110 89 2014204 82 2011024 81 2014109 39 2011010 11 2011010 25 2011025 78 2011026 42 2006148 35 2014106 16 2011019 20 2011027 38 2006148 48 2011018 62 2011007 91 2014204 59 2011021 29 2014111 41 2011024 66 2013279 23 20011731 2011026 82 2013281 48 2014108 99 2011024 88 2014107 73 Subject_id: 201327986 2014111 62 2013281 00 2006148 03 2011018 94 2006141 93 2015010 81 2014109 60 2011016 52 2014107 38 2011022 85 2006155 93 2011011 85 2011023 72 2013281 61 2014205 43 2015010 64 2011010 48 2014108 75 2014107 85 2014111 79 2011006 45 2011009 88 2011021 22 2014205 05 2011006 50 2011007 18 2011022 21 2014205 30 2011012 46 2011019 04 2014110 58 2014106 30 2011012 39 2011026 76 2007237 01 2011023 17 2011024 60 2011017 86 2011008 56 2014205 08 2011024 45 2014107 46 2013280 26 2011022 37 2011023 63 2014204 43 2011007 11 2011024 55 2006146 40 2014205 00 2011007 45 2014108 45 2011010 86 2014110 13 2003085 12 2011022 55 2014106 98 2011021 46 2014107 63 2011009 55 2002113 18 2011021 89 2007237 24 2011009 80 2011008 20 2011025 28 2011009 85 2011010 61 19982780 2014106 14 2011006 96 2006142 08 2011018 96 2014106 78 2014110 16 2011026 53 2011009 87 2014107 03 2013280 36 2011017 95 19982793 2011009 81 2011023 62 2007237 34 2011024 29 2014106 41 2011010 32 2014204 39 2014109 38 2007237 36 2011008 77 2011009 84 2014106 12 2011016 50 2014108 21 2007233 11 2011020 59 2011009 39 2006148 37 2006146 50 2011023 50 2014204 25 2011009 44 2014109 29 20027368 2011020 77 2015010 88 2011006 56 2011022 52 2011024 94 2015010 60 2011027 41 2011012 90 2011016 72 2013279 60 19890060 2011025 15 2014106 42 2011018 52 2011026 86 2011022 01 2013278 84 2011019 22 2014111 30 2011022 67 2014111 61 2006150 80 2014106 19 19982788 2011021 03 2006142 03 2014107 42 2014108 06 2011009 67 2011017 05 2011019 50 2014107 47 2011010 78 2011026 00 2013280 58 2011021 63 2014109 72 2014108 76 2013279 82 2014110 36 2014106 92 2011022 11 2011021 78 2011012 72 2014107 48 2004052 93 2011010 90 2006146 08 2011009 12 2014108 92 2013281 16 2014108 93 2013280 03 2011015 62 2011015 57 2007238 22 2011027 14 2014111 56 2014109 33 2014108 13 2011021 16 20015548 2014111 52 2004024 98 2011021 31 2013280 67 2011021 41 2011025 74 2004057 59 2014109 73 2013279 73 2003085 24 2011019 60 20000236 2014110 88 2011027 11 19982154 2011010 75 2011007 52 2011026 94 20020182 2007237 28 2011019 07 2015010 84 2011023 45 2014109 03 2002113 48 2014107 81 2011025 98 2011024 72 2011010 62 2014106 64 2014109 49 2006148 65 2014109 40 2014109 19 2014109 25 2011008 91 2014108 89 2011021 84 2011017 88 2011022 03 2014111 13 2014109 54 2013279 17 N : 61 N : 29 2011007 26 2011025 95 2011024 41 2013279 16 2014111 36 2011022 60 2006145 91 2011026 32 2014204 46 2011018 91 2006148 63 2006146 62 2006140 85 19994695 2011023 34 2011021 73 2004064 07 2011025 77 2014106 79 2011006 57 2007235 34 2011018 53 2014108 08 2014111 70 2011009 20 2011006 34 19982749 2003085 55 2011025 60 2011008 57 2006138 69 2011008 44 2011026 95 2011020 58 2015010 89 2007237 44 2013281 32 19982775 2011018 72 CTRL CTRL 2011027 03 2006142 58 2014108 14 2011017 10 2011010 74 2013279 21 2011010 27 2014106 38 2011024 23 2014107 13 2013279 89 2014107 41 2007233 45 2011020 54 2014109 84 2007231 87 2006155 41 E: 98 E: 45 2014110 25 2014204 97 2006155 64 2002113 02 2006141 71 2014110 70 2013278 87 2006142 15 2011024 56 2003085 19 19994628 2011008 26 2011009 97 2011023 65 2011020 21 2006142 89 2011023 40 19982808 2014108 67 2011012 81 2011023 29 2011019 13 2014108 12 2014109 89 2011016 19 2011007 49 2013279 40 2006148 45 2006146 69 2011022 02 2015010 79 2014111 01 2013280 65 2011021 86 2014110 81 2011017 81 2011009 65 2011008 90 2011022 83 2011024 00 2011025 29 2014106 13 2014109 22 2013280 06 2014111 42 2011010 26 2011020 56 2007233 49 2011009 99 2014110 00 2014204 38 2011009 04 2011007 68 2013280 28 2011021 59 2013281 27 2013280 71 P: 1 P: 1 2011022 54 2011012 45 2014110 28 2011022 42 2011007 25 2011024 13 2013280 85 2011019 79 2013280 61 2013281 29 2011020 67 2014110 12 2011021 67 2011023 11 2011023 42 2011014 00 2014106 33 2011026 96 2013281 59 2011023 81 2006141 27 2014106 81 2014110 80 2011010 14 2014111 23 2013280 21 2011015 35 19995468 2011009 33 2013281 06 2011022 25 2014110 97 19960638 2006140 25 2006142 80 2011024 79 2013279 71 2006142 18 2011016 39 2011006 78 2011019 25 2011011 79 2011019 75 2014111 48 2014108 39 2011018 93 2011007 27 2011007 84 2013279 70 2011025 73 2014205 11 2011025 14 19994453 2011026 50 2011022 76 2011013 00 2011024 73 2013281 21 2014108 18 2013279 86 2007232 65 2011018 45 2011026 17 2011012 68 2014111 12 2011025 53 2014204 31 2014205 36 2014111 19 2011020 55 2014107 14 2014108 00 2011006 46 20020098 2007237 32 2014109 14 2006155 63 2011006 94 2011015 23 2011010 46 2013280 14 2013279 30 2011025 81 2011022 12 2014109 62 2015010 74 20011751 2011017 42 2014107 60 2014111 54 2013281 01 2014204 50 2013280 91 2006146 55 2011015 14 20020087 2014108 53 20000750 2011016 44 2011023 86 19994223 2011007 37 2011022 98 2011019 77 2006142 63 2011007 41 2014109 31 19994688 2011025 16 2014107 83 2011008 65 2011024 49 2011026 93 2011023 23 2011019 71 20020251 2011014 47 2011006 68 2013280 05 2007233 08 2011010 68 2011025 02 2006146 29 2004057 62 2011024 42 2014110 52 2011021 06 2014111 75 2013279 77 2014110 76 2011026 97 2011022 20 2011024 65 2011019 05 2011018 55 2013281 11 2011006 75 2014110 61 2014109 00 2011008 79 2011009 64 2011016 41 2013279 65 2011015 64 2011024 03 2011022 06 2011007 77 2014108 37 2011025 76 2011008 14 2015010 87 2011023 51 2011009 38 2007237 27 2011021 68 2011016 73 2014110 74 2011027 10 2013279 51 2011009 96 2003085 48 2013281 65 2007232 50 2011010 85 2011020 64 2011018 85 2011023 15 2011025 83 2014111 22 2011016 48 2006140 57 2014111 04 2006141 68 36 2013279 2011006 35 2011006 72 2011010 77 2011008 94 2011023 20 2011025 50 19994486 2011019 10 2014106 66 2011013 32 2007237 42 2011012 62 2013280 99 2011025 70 2014111 47 2011023 28 2014108 72 2011012 23 2011009 27 2013278 72 2013278 94 2011025 88 2011026 79 2011021 48 2013280 31 20004964 19890061 2014107 65 2013280 41 2013280 68 2011018 21 2011017 71 2011021 61 20020219 2011027 06 2014107 98 20020200 2002112 99 2014110 32 2015010 76 2011023 73 2013280 40 2011024 86 2011024 36 2014110 22 2006142 55 2011026 56 2011007 62 2014205 41 2014204 54 2006140 60 2014110 29 2011010 80 2013280 70 2006142 52 20015488 2011025 19 2013281 63 2011017 90 20027301 2011024 46 2013280 66 2013280 12 2011019 82 2011020 16 2011006 36 2014108 43 2014106 67 2011020 85 2011025 17 2011022 10 2006140 49 Subject_id: 20020214 Subject_id: 200723721 2011019 59 2014106 80 2014110 08 2007233 02 20011741 2011018 39 2015010 77 2011007 46 2011022 79 2011019 80 2006142 77 2014110 14 2006142 91 2013281 13 2011011 06 2006146 04 2006142 49 2011017 87 2011021 53 20000122 2011009 94 2011021 43 2011026 77 2003085 53 2011019 32 2014108 59 2011007 21 2006145 79 2011014 33 2011022 41 2013279 53 2011025 42 2011013 29 2011021 99 2013280 16 2011017 84 2014204 58 2011027 17 2014205 37 2003085 44 2006155 94 2011014 72 2011021 42 2015010 93 2006141 59 2014205 29 2014109 09 2014106 68 2011016 67 2013281 33 2015010 73 2006142 21 2011023 48 2014107 89 2013278 95 20015405 2011012 14 20011716 2013278 83 2011014 55 2014204 80 2011025 55 2014110 78 2014110 56 2011013 89 2014111 39 2014204 83 2013279 59 19881755 2014110 46 2014204 85 2006147 48 2011025 97 2011009 02 2014204 60 2011010 99 2014110 82 2011022 27 2011027 37 2014205 23 20020242 2011011 81 2011019 70 20027317 2011013 15 2013280 90 20020490 2004025 42 2006142 40 2011020 74 2011009 14 2011009 13 2014108 11 2014205 21 2011025 71 2011020 17 2011018 78 2011018 87 2014205 34 20020109 2011023 99 2013280 48 2004051 22 2011007 61 2006146 80 2011009 91 2014109 88 2011020 84 2011021 58 2011025 01 2014111 77 2006147 42 2013279 80 2011010 33 19982745 2011016 71 2011022 64 2011026 61 2011011 27 2011020 14 2013280 92 2011011 98 2013279 67 2014107 35 2011026 12 2013280 83 2006146 12 20015408 2014108 03 2006141 69 2014108 31 2011008 84 2014204 61 2014108 49 2014107 32 2011009 41 2011017 67 2011021 49 2011014 79 2011007 15 2011011 21 2011022 68 2011014 01 2014106 48 2006140 89 2013281 15 2011021 71 2013280 10 2011023 57 2011016 18 2011011 38 2014109 96 2011021 40 2011023 26 19982713 2014109 11 2011007 64 2011019 88 19994707 2011022 18 2011008 70 2015010 85 2014108 30 2014109 45 2011027 34 2013280 88 2013280 86 2011020 86 2011026 38 2014106 91 2011019 81 2014111 09 2013281 18 2014108 91 2011016 10 2011022 32 2011026 40 2011025 87 2014107 05 2011025 37 2006155 83 2011020 09 2014111 82 2011020 98 2014204 93 2014109 93 2014111 43 2011019 97 2006155 89 2007236 62 2007237 29 2006141 66 2014109 77 2011011 04 2011010 12 2011010 17 2006140 90 2014204 94 2011011 36 2006142 69 2011023 83 2011026 91 2011024 78 2007233 50 2011021 14 2011018 95 2006155 74 29 2013279 2011019 31 2011025 45 2006146 33 2013281 14 2014111 58 2011023 30 2014106 32 20020088 2014106 40 2011006 43 2014106 94 2013281 43 2011014 78 2011008 58 19982145 2013280 45 2011025 41 2013279 32 2013281 64 2013280 47 2014204 88 2013280 77 2011024 63 2013279 95 2011021 87 2014109 32 2011009 00 2006142 82 2014204 33 2014204 20 2011021 47 2011020 11 2014110 51 2011009 82 T cell CNS 2011009 34 2011025 85 2006142 61 2011010 10 2006141 74 2006152 26 2011026 70 2013279 37 2011018 40 2011020 29 2014110 75 2014110 21 2014110 73 2011016 27 2011008 74 2013279 93 2006145 14 2014204 40 2011013 31 2013279 11 2014108 87 2011018 92 2011019 16 2011014 73 2013278 91 2011015 05 2011009 47 2011024 20 2006151 40 2011025 23 2011024 34 2006146 13 2011019 21 2014111 11 2006148 47 2011019 91 2011023 13 2011025 33 2004054 71 2007240 69 2014106 44 2014108 50 2011022 78 20015673 2011027 04 2006140 66 2014107 19 2011021 55 2014109 41 2013279 69 2014107 15 2011012 09 19920399 2014205 04 2007237 12 2014106 39 2011022 07 2014107 80 2014107 57 2014204 52 2014106 43 2007236 61 2011024 71 2014204 45 2014110 35 2011010 95 2014108 68 2011024 57 94 2011007 20020028 2011010 51 19994456 2011022 28 2011013 08 2011012 55 2011023 96 2011007 53 2011017 12 2013279 49 2011016 20 2014109 53 2006142 66 2013279 52 20004903 2013278 34 2011023 91 2011027 09 2013278 98 2014109 95 2011009 49 2011020 23 2011019 92 2011010 97 2011022 38 2006142 79 2006142 97 2011026 21 20000518 2014109 21 2014108 54 20005081 2011026 72 2011019 66 2014111 07 2014108 38 2011023 90 2011019 06 2006146 16 2014204 96 2011023 04 2004050 38 2014110 45 2011025 38 2014107 82 2011027 00 2014108 22 2014106 61 2013281 40 2011008 46 2011024 16 2014204 98 2011011 67 2014109 97 2011023 31 2011020 65 2011024 84 2011022 70 2013281 23 2011023 12 2013281 03 19994542 19961103 2014110 55 2011026 60 2014111 60 2014108 74 2014106 54 2011021 79 2011025 92 2014110 26 2011027 07 2011023 58 2011009 56 2013279 85 2014108 97 2011007 73 2011010 30 2014111 20 2014109 17 2014108 96 2007232 96 19994462 2014107 08 2006146 67 2013281 38 2006145 59 2014111 34 2014107 61 2013279 88 2011022 84 2015010 72 2011008 61 2014111 16 2013281 60 2011023 82 2014109 42 2011015 13 2011012 71 2014107 51 2014108 77 2011023 36 19982805 2013281 07 2014110 07 2011008 54 2014205 12 2011010 20 2011024 38 2014108 19 2013279 43 2013280 72 2011019 24 2013279 54 2007237 45 2013280 08 2014106 53 2011018 58 2011026 55 20020396 2013280 27 2011014 32 2011021 09 2011012 32 2013279 01 2007232 82 2011022 36 2011016 45 2011009 98 2013279 10 19982757 2011019 87 2014109 06 2014109 63 2011025 89 2011020 70 2014106 75 2011016 25 2014108 33 2011015 50 19994502 2011020 05 2011026 06 2013279 87 2011025 34 2011022 74 2014110 68 2014110 60 20014823 2014109 57 2011008 69 19994167 2014106 63 2011010 54 2011023 64 2011021 20 2014107 97 2011010 09 2003085 49 43 2011009 2014108 62 2014107 33 2006148 64 2014111 29 2014108 17 2011026 78 19994318 2011011 19 20011781 2011013 18 2011010 57 2011007 86 2007237 31 2011024 37 2014107 99 2014108 41 62 2011019 2011023 60 2011010 31 2011007 78 2011022 75 2011010 59 2011022 59 2011022 30 2011022 91 2014106 58 20020256 2011016 70 19994459 2011019 94 2007236 95 N : 99 N : 59 2014107 36 2006148 62 2011009 79 2004009 20 2013280 37 2011025 25 2014108 95 2011026 36 2011009 46 2011014 60 2011020 60 2014109 99 2006145 42 2014109 90 2014205 28 2013280 15 2011009 21 2014111 55 2011020 33 2014108 24 2014204 17 2011020 46 19982750 2011009 32 2014109 91 2006145 74 2006155 75 2011018 49 2011006 67 2011021 64 2011022 26 2014204 69 2011024 01 2013278 97 2007233 68 2011019 29 2013279 98 2011016 42 2011023 94 2011024 87 2011021 74 2014107 92 2011021 85 2011017 19 2014111 59 2011009 26 2011025 39 2011020 50 2013278 73 2011014 31 2014106 36 2013278 82 2007237 03 2011023 77 2014110 53 2011024 61 E: 172 E: 91 2014106 49 2006155 61 2011023 79 2011017 04 2011023 16 2014109 55 2011025 35 2013280 62 2007237 15 2014108 84 CASE CASE 2011024 28 2014106 93 2014109 02 2011018 81 2007240 18 2011009 37 2011022 65 2011008 92 2014111 27 2011007 04 2004051 16 2015010 70 2011020 37 2014111 45 2006143 00 2013280 75 2011008 36 2014110 99 2011010 41 2013278 41 20020205 2013281 57 2013279 79 2014106 62 2011024 32 2014109 82 2011019 61 2014107 86 2014205 27 2006142 84 20014829 2014107 18 2011014 46 2014111 17 2014111 02 2014110 41 2011019 90 2014205 20 2011009 61 2007234 96 20020211 2015010 71 19994372 2014106 73 2006151 44 2006141 57 2014108 15 2013278 39 19982758 2011024 35 2011026 24 2011023 41 2011026 30 2013278 76 2013281 55 2014108 36 2006148 04 2014107 00 62 2013279 2014204 62 2014108 32 P : 94 P : 85 2007237 22 2013279 18 2014108 25 2006137 22 2014110 87 2013281 08 2011023 24 2013278 31 2014108 20 2011010 56 2014106 89 2014106 46 2014109 75 2011022 92 2011012 94 2006155 69 2014106 34 2011025 47 2011023 74 2006142 74 2011009 40 2011021 11 2006145 41 2013281 25 2006141 60 2011026 74 2011020 48 2011022 88 2014110 85 2006151 45 2013279 50 2014205 35 2011023 61 2007237 39 2014205 45 2011009 18 2014108 23 2011025 08 2013280 35 19994439 19995499 19994519 2014110 64 2011007 83 2011024 75 2014109 56 2014107 44 2011014 99 2011021 10 2011010 28 2011024 97 2014111 15 2013279 28 2014107 69 2006136 08 2014108 55 2014107 62 2014110 83 2011007 99 2011024 76 2006141 24 2011023 67 2014108 88 2011026 35 2014108 51 2013278 80 2013281 31 2014107 52 2005044 87 2011021 02 2014110 79 2011026 92 2014109 34 2011022 63 2014106 70 2011007 69 2011027 05 2011017 34 2013278 47 2014110 39 2011020 63 19890304 2013278 77 2013279 72 2006146 22 2013280 01 2011022 57 2011017 92 2006142 76 20020230 2014107 30 2014106 22 2011008 59 19994380 2007237 16 2014110 86 2011018 51 2011014 80 2006142 29 2014110 30 2011010 16 2011025 59 2011012 82 2011026 54 2011012 33 20000089 2014108 10 2011023 53 2014111 49 2011022 48 2006151 41 2014106 29 2011019 72 2011022 34 2013279 31 2007233 51 2006148 16 2011025 52 2014109 46 2011022 33 2014109 44 2011025 18 2011019 48 2006141 62 2011020 45 2011024 30 2013279 42 2013279 07 2011017 96 2014107 26 19994319 2011018 74 2003085 39 2011009 07 2015010 83 2011027 39 2014108 58 2011016 22 2011010 45 2011024 50 2011018 61 2007233 21 2011009 60 2011007 70 2011025 11 2007237 06 2006146 11 2011008 55 2011018 64 2006142 50 2014108 35 2014111 37 2011009 52 2004055 61 2011022 96 2011020 68 2011024 44 2011007 10 2011021 07 2011009 24 2011013 96 2007233 38 2015010 59 2013278 70 2011025 43 2011020 94 2011010 02 2014108 57 2011014 16 2011010 24 2011023 71 2011008 49 2011018 88 2013280 44 2011010 49 2011006 86 2011018 76 2014111 57 2007237 25 2011025 64 2013278 71 2011010 71 19994327 2002114 08 2011026 90 2011010 37 2013281 22 2011007 58 2011015 30 2011026 64 20000085 20015510 2014204 47 2011010 72 2011018 42 2006141 31 2013279 14 2011010 79 2005190 53 2007242 15 2014107 10 2014109 26 2011023 18 2011007 34 2011025 36 2011026 20 2013279 83 2011021 70 2006148 34 2011007 56 2013280 34 2013281 10 2011007 96 2014205 31 2011014 98 2014110 19 2014111 38 2011018 65 2014111 68 2006140 84 2006155 88 2011024 89 2011012 99 2011018 75 20020512 2011017 07 2011022 16 2014111 00 2013279 90 2011008 73 2011011 00 2011026 02 2011008 10 2014204 91 2011019 27 2013279 33 2011018 90 2007240 10 19982785 2011009 08 20020282 2011020 69 2011021 83 2013279 08 Subject_id: 201102205 2011021 18 2013280 30 2015010 57 2013279 91 2011017 02 2011006 74 2014109 20 2014108 42 2014106 17 2011009 35 2011016 51 2011020 42 2011008 33 2011020 18 2011020 82 2011026 43 2013280 09 2011010 87 2011006 90 2011021 57 2011008 88 2014108 16 2011018 41 2014205 25 2014108 63 2011007 93 2014106 74 2011027 21 2011014 76 Subject_id: 201100870 2013281 54 2011024 40 2006141 98 2011008 52 2011018 44 2011010 55 2011026 01 2013280 22 2011026 68 2011007 00 2014110 11 20014825 20011739 2013279 41 2006142 65 2011008 64 2011007 42 2011006 37 2013279 66 2011010 22 2014110 63 2011019 00 2011025 49 2011026 09 19982789 2013281 41 2011026 89 2011026 37 2014107 11 2011023 44 2013281 62 2014107 49 2011006 69 2013281 36 2011020 04 91 2011024 2006140 88 2014110 42 2011021 25 2011008 53 2011019 14 2006145 85 2015010 66 2011009 77 2013281 58 2006140 95 2014108 52 2006140 63 2014111 50 2011020 30 2011026 57 2014204 42 2011023 19 2011010 42 2014110 91 2011010 70 2011018 97 2006141 29 2006148 61 2014109 61 2013279 22 2006155 68 2011026 65 2011019 56 2014107 02 2014109 85 2006148 51 2011011 87 2011007 89 2011006 49 2013279 46 2011022 08 19982792 2014205 16 2011010 65 2014107 70 2014106 50 2011027 43 2011025 99 2011024 82 2014109 71 2011015 48 2011010 91 2011022 22 2011010 92 2013281 12 2011020 03 2011019 42 2011023 03 2014109 18 2013280 50 2013280 94 2011009 48 2013280 89 2011016 01 2011026 27 2014111 51 2014108 65 2004007 22 2011022 19 2011026 34 2014110 23 2014110 66 2011025 80 2011019 47 2014107 54 2014111 73 2011008 13 2006141 79 2004008 87 2014109 23 2011009 17 2011027 16 2011022 17 2011008 68 2014106 18 2011013 33 2014107 34 2011021 27 2014109 28 2013281 04 2014204 51 2011009 45 2011010 50 2011020 39 2014110 57 2011022 82 2014204 22 2004049 51 2014106 85 2014109 30 2014107 37 2011009 58 2011022 72 2007240 11 2011013 42 2013279 39 2011020 88 2014107 22 2011008 07 2011010 84 2011007 87 2014205 26 2014107 79 2014108 78 2011007 71 2015010 68 19994450 2004009 47 2011026 47 20020023 2006155 70 2011007 54 2007237 40 2006141 00 2013278 99 2014108 40 2013279 00 2014110 34 2015010 69 2006142 73 2011026 52 2011018 67 2011007 79 2011010 00 2014106 55 2014109 66 2006142 56 2013280 57 2011017 27 2006141 28 2013280 53 2011020 51 2011025 00 2013280 00 2011022 69 2011023 78 2011019 28 2011007 59 2011019 74 2011024 26 2011025 96 2011017 06 2011018 83 2011010 47 20020435 19994510 2013281 42 2011019 30 2007236 73 2011008 38 2014106 59 2011018 13 2015010 75 2011009 19 2011022 39 2014107 23 2011012 56 2014108 46 2011017 80 2013280 24 2011011 05 2015010 86 2011006 95 2013280 73 2011019 40 2011026 80 2011026 11 2011024 25 2011020 78 2006136 38 2011010 29 2011006 98 2011011 17 2013279 75 2011024 15 2014204 63 2011019 76 2013279 27 2011020 92 2013278 85 2006142 25 19960626 2014107 28 2014109 15 2013278 93 2011010 98 2011021 92 2011025 72 2006147 73 2011019 23 2011020 90 2007240 27 2011010 60 2011019 38 2011018 70 2013278 90 2011009 83 2011026 51 2007237 23 2011014 74 2011021 95 2011020 73 2014108 94 2011022 66 2013280 79 2011022 50 2011020 24 2011008 24 2014110 65 2011027 08 2011018 60 2013279 94 2011018 86 2011023 75 2011021 05 2014109 69 2006145 11 2006148 18 2006141 82 2011014 83 2014110 93 2014110 98 2013281 51 2011008 19 2006140 42 2011010 66 2013279 38 19994358 2007234 05 2011024 85 2011021 36 2013280 39 98 2006142 20020265 2014106 28 2011024 10 2011026 18 2011019 67 2014109 10 2006142 86 2007233 31 2011009 74 2011026 05 2011022 05 2014205 15 2011024 51 2014204 53 2014110 05 2013280 32 2014106 26 89 2011022 2011007 43 2014107 72 2011007 24 2013280 59 2006146 03 2014205 02 2011009 89 20000272 2014106 47 19982732 2011008 85 2011027 19 2011009 25 2011014 39 2013279 81 2011013 64 2007237 04 N : 53 N: 9 2011024 05 20020189 2011012 50 2006146 14 2014109 16 2014204 75 2011007 65 2013281 37 2011010 94 2013280 84 2014108 61 2011008 63 20020161 2014111 03 2011025 21 2011023 54 2014110 15 2014110 59 2011026 81 2014107 16 2011008 12 2011015 21 20011714 2015010 67 2014109 59 2011021 76 2013280 19 2011026 29 2014106 95 2011021 93 2011008 83 2011019 39 2011007 36 2011019 98 2011017 89 2011007 72 2011025 32 2014106 37 2013279 63 2007242 85 CTRL CTRL 2006146 41 2011010 13 19982725 2011009 66 2011019 89 2013281 46 2011024 17 2011008 43 2011011 16 2011019 35 2013281 19 2006146 48 2011024 27 2011026 15 2006153 71 2014111 35 2011021 88 E: 60 E: 8 2011016 43 2011026 22 2014109 01 2011010 58 2014107 27 2006145 84 2006146 63 2011008 76 2014109 67 2011022 87 2011023 98 2011020 93 20020623 2011027 15 2011009 05 2014204 89 2013281 35 2014109 79 20020176 2013279 15 2011024 14 2006142 94 2015010 94 2013279 74 2006145 48 2014109 80 2011008 16 2011023 00 2013279 34 2011025 94 52 2013281 2007237 43 2011007 44 2014110 69 2006147 85 2014110 96 2013281 34 2014108 34 2011017 03 2011018 18 2014108 85 2014110 43 2014106 83 2011019 84 2011025 58 2011007 35 2013279 25 2013280 18 2011010 53 2011023 68 2011022 15 2013279 96 2003085 57 2014106 96 2006146 53 2013279 99 P: 1 P: 1 2011020 81 2011019 44 2014106 20 2007237 21 2011017 09 2011011 01 2006141 63 2011018 73 2011009 95 20027304 2006142 47 2015010 80 2014110 44 2011019 53 2013281 68 2011009 86 2011025 22 2015010 63 2011021 13 20020214 2014108 80 2014107 17 2013279 84 2014110 02 2011025 91 2004056 78 20000164 2011020 13 2011019 02 2011017 93 2013281 17 2014110 37 2014107 56 2014109 83 2011026 31 2011021 80 2011010 63 2011008 18 2014107 74 2014109 48 2014205 09 2014107 20 2014111 53 2011019 93 2014205 46 2014110 01 2013280 63 2014106 90 2014106 51 2014107 91 2013281 30 2011026 13 2014205 33 2011023 88 2014108 29 2013280 46 2011021 97 19994566 2007237 33 2011013 77 2013280 96 2011020 71 2013280 04 19994465 2014110 49 2013280 25 2014204 27 2011009 51 2011027 44 2014107 59 2014109 70 2011019 18 2014108 79 2011026 33 2014108 07 2011020 89 19994442 2006142 43 2014107 53 2014109 51 2014109 74 2014204 24 2014107 75 2011020 66 2011021 38 2011024 31 2011008 03 2015010 82 2014106 52 2011008 30 2011026 46 2006148 58 2014106 27 2014205 38 2011018 98 2011021 60 2014106 65 2011024 68 2011011 10 2014204 95 2014204 36 2015010 58 2014108 60 2014110 84 2011024 83 2011007 38 2014107 31 2013280 17 2014205 10 19982798 2014108 81 2007236 89 20020170 2013279 45 2014108 28 2011008 21 2013279 24 2011006 83 19982800 2011025 13 2014107 93 2011007 05 2011015 38 2011008 28 2014204 99 2014110 24 2011013 09 2014109 65 2007237 19 2011025 51 19982731 B C M T Subject_id: 20020214 Subject_id: 201328146 1 200 Fig. 5 Select case-control intra-individual MS-risk networks. a Number of edges in the largest connected component (LCC) of the network generated among proteins (genes) with high PRE (>25th percentile) in 2370 patients and 412 healthy controls (GW_r2 > 0.5). Each row represents a subject, each column represents a cell type (B: B cell; T: T cell; M: monocyte; c: CNS). The leftmost column indicates subject status (red: cases; green: controls). b Representative sub-networks from subjects at the extremes of the distribution for E-LCC for each cell type. For each network, the number of nodes (N), edges (E), and percentile relative to all subjects (P) is indicated. The intensity of node color is proportional to the PRE of each gene in the corresponding cell type variant(s) but also provides an interpretable model of risk in a anatomical regions as available in the ENCODE and REP data- cell/tissue specific manner. While the PRE scores computed here sets. Although all derived from CNS tissue, it is highly likely that are significantly correlated with actual gene expression from the different regulatory mechanisms are at play in each cell type (in corresponding cells, the correlation is partial, thus underscoring a some cases resulting in drastically different expression patterns), potential limitation in our approach, and of available data. thus somehow confounding the overall CNS signature computed However, the statistically significant results obtained for genes here. Thus, the detected effect of MS-associations which map to and pathways known to be involved in MS further validates this the CNS could represent the lower boundary of a more wide- approach. Indeed, the models presented here are consistent with spread phenomenon. A more detailed CNS-specific data set MS genetic risk being driven by the long-term alteration of cel- containing genome-wide regulatory element information might lular pathways primarily in monocytes, but also in the B and T be needed to address this question in larger detail. cell arms of the human adaptive immune response. The smaller In the last few years, several post-GWAS pathway approaches but not negligible contribution of CNS pathways to MS risk is in such as DEPICT37, FUMA38 and PASCAL39 have been proposed agreement with our previous analysis17, which identified the and utilized to interpret and integrate summary statistics into a monocyte/macrophage/microglia axis as a key player in directing biologically meaningful model. While sharing some of the basic the autoimmune process to the CNS. However, an important characteristics of previous approaches, our method features a set caveat, particularly affecting results for this compartment, must of unique properties, most notably the introduction of data- be taken into account: for this analysis, the CNS group was driven regulatory effects of associated variants (and those in LD), composed of a heterogeneous ensemble of purified primary cells, the ability to create cell-specific networks, and the computing of established cell lines and dissected specimens from specific individual disease burden maps (Supplementary Table 1). Given NATURE COMMUNICATIONS | (2019)10:2236 | https://doi.org/10.1038/s41467-019-09773-y | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09773-y B cell T cell Monocyte CNS Subject_id: 201327986 N : 110 N : 104 N : 116 N : 57 E: 190 E: 164 E: 201 E: 88 P : 99 P : 47 P : 49 P : 66 Subject_id: 201101471 N : 122 N : 102 N : 107 N : 58 E: 134 E: 219 E: 197 E: 84 P : 52 P : 98 P : 65 P : 53 Subject_id: 201102205 N : 77 N : 99 N : 114 N : 46 E: 124 E: 172 E: 249 E: 69 P : 30 P : 60 P : 94 P : 28 Subject_id: 201101897 N : 91 N : 122 N : 118 N : 78 E: 121 E: 189 E: 206 E: 118 P : 129 P : 51 P : 75 P : 99 Fig. 6 Heterogeneity in intraindividual MS-risk networks Intraindividual cell-specific networks of four representative MS subjects showing heterogeneity of risk across all cell types. a Cell specific risk networks for subject_id: 201327986. b Cell specific risk networks for subject_id: 201101471. c Cell specific risk networks for subject_id: 201102205. d Cell specific risk networks for subject_id: 201101897. For each subject, the most connected risk network (number of edges in the highest percentile across all subjects) is highlighted within a colored box. For each network, the number of nodes (N), edges (E), and percentile relative to all subjects (P) is indicated. The intensity of node color is proportional to the PRE of each gene in the corresponding cell type. M: monocyte, green; T: T cells, red; B: B cells, blue, C: CNS, yellow that each method produces a different output, it is not possible to speculate that in the near future this information could also be directly compare these approaches. However, the first and argu- used as the basis to develop individualized risk scores, or to derive ably most important step in all these tools (including ours) is to personalized approaches to therapy. For example, a subject with compute the SNP to gene values. A basic comparison of these high B cell genetic risk may be a good candidate for B cell three tools shows that the exact same genes were prioritized for depletion therapies, while a subject with a high T cell risk may almost half of all non-HLA associated loci (97/200). A closer look benefit the most from immunomodulatory drugs that target revealed that our method was the only one that called at least one T cell function or migration into the CNS. gene per associated locus, and produced gene prioritization sets This cell-specific pathway approach can be extended to any with the least ambiguities (Supplementary Table 2 and Supple- set of SNPs of interest in any condition at both population mentary Fig. 5). (summary) and individual (genotype) levels. Recent evidence has emerged that polygenic risk scores for schizophrenia associate with therapeutic response to Lithium- Methods based therapies40. Similar approaches are being tested for other Predicted regulatory effects (PRE). Genome-wide regulatory elements from psychiatric, oncological and cardiovascular diseases41–45. The ENCODE and REP were collected from regulomeDB46 (which contains more than observation that some risk variants only affect expression of a 400 million genomic regulatory features collected from 400 cell and tissues; Sup- given gene in one cell type but not in others, may at least in part, plementary Table 3) for all non-MHC independent effects (SNPs)17. Specifically, single nucleotide polymorphisms (SNP) corresponding to all non-MHC GW (n = underlie the observed clinical heterogeneity in the MS population. 200; Supplementary Data 33), SR (n = 416; Supplementary Data 34) and NR (n = Thus, when this approach is implemented at the individual level, 3695; Supplementary Data 35) were extracted for analysis GW, SR, and NR as specific risk profiles can be built for each subject with MS. We defined previously17). The 200 GW effects were distributed in 156 unique regions 8 NATURE COMMUNICATIONS | (2019)10:2236 | https://doi.org/10.1038/s41467-019-09773-y | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09773-y ARTICLE (44 regions contained multiple independent effects). Similarly, the 416 SR effects replication and reached genome-wide significant in the final combined analysis were distributed in 354 unique regions (62 regions contained multiple independent (GW), those that showed statistically significant evidence of replication but did not effects) and the 3695 NR effects were distributed among 1,883 unique regions reach genomewide significance in the final combined analysis (SR) and those (1812 regions contained multiple independent affects). Three sets of SNPs were failing to show statistically significant evidence of replication (NR). For each cell created for each region according to their r2 with their corresponding main effect type, the number of nodes and edges of each subnetwork and that of its largest (r2 ≥ 0.8, r2 ≥ 0.5, and r2 ≥ 0.1). connected component were computed. The statistical significances were computed A python tool was written to automatically fetch data from RegulomeDB for by comparison against a background distribution of 10,000 networks of equal size these SNPs in all three lists (totaling 538,826 SNPs). Similarly, data for SNPs in sampled randomly from the same PPI. different levels of LD (r2 > 0.8, r2 > 0.5, and r2 > 0.1) with each primary effect were also retrieved using chromosomal positions. The main analysis was performed Cell-specific transcriptomes. This work was approved by the Institutional Review using r2 ≥ 0.5 whereas the other sets were only used for the sensitivity analysis. In Board at the University of California San Francisco (IRB# 10-00104). PBMCs were total, 538,826 SNPs were included in the analysis. obtained from 25 individuals by Ficoll method using Vacutainer CPT tube (BD To investigate the effect of SNPs across different cell types and to assess which Biosciences). Subjects were consented according to institutional (UCSF) review gene has most potential of being regulated across various cell types, we grouped the board (IRB) guidelines. Three different cell subsets (CD4+ and CD8+ T cells, CD14 cell types present in ENCODE (Supplementary Table 4) and REP (Supplementary + monocyte) were sorted into RLT buffer using a MoFlo Astrios cell sorter Table 5) into four major cell types (buckets). Specifically, these were B cells, T cells, CNS (central nervous system), and M (monocytes). T cell subsets (Th1, Th2, Th17, (Beckman Coulter). Helper T cells were defined as CD3+CD19−CD4+, cytotoxic and Treg) were also analyzed as a separate group. We also built a dataset from lung T-cell were CD3+CD19−CD8+, and monocytes were sorted as CD14+ cells. Total (L, a cell type/tissue not considered to play a major role in MS susceptibility) as a RNA was isolated from sorted cell subsets using RNeasy Mini kit (Qiagen) and control. Cancer cell lines were excluded for this analysis. assessed RNA quality using Agilent 2100 Bioanalyzer (Agilent Technologies). 3′ Regulatory elements were grouped into two major classes: PEX (promoter/ mRNA-Seq libraries for all cell subsets were prepared from 100 ng total RNA using enhancer/activator) and R (repressors/inactivators). Cell or tissue of origin was QuantSeq kit (Lexogen) according to the manufacturer’s instructions and recorded for each regulatory feature, and cell-specific information was grouped into sequenced 50-bp single-end on the HiSeq 4000 (Illumina). Sequence reads were three main cell types (B cells, T cells, monocytes) and one tissue (CNS) that are of mapped to the human genome reference (GRCh38) with Gencode annotation (r26) interest in MS. In total, 25 brain regions were considered (15 from ENCODE and using STAR aligner50. Reads were normalized by median of ratios using the DEseq2 10 from REP). In addition, T cell subsets deemed relevant in the pathogenesis of package51. The R function featureCounts was used to obtain gene-level read MS (Th1, Th2, Th17, and Treg) were also analyzed separately. Primary cells and counts52. cell lines from a tissue not known to be involved in MS (lung, L) were included as We selected overlapping genes between RNA-Seq gene counts and PRE scores, control. In addition, eQTL data for T cells and monocytes from the IMMVAR and Pearson’s correlation test was performed using the cor.test function in R. The project47 were integrated into the PRE computations. The HTML data from the significance of the correlation was confirmed by permutation testing (n = 1000). scrapped output was parsed to populate data present in regulomeDB tables. A master table was then compiled with each field of regulomeDB data for all Data availability 538,826 SNPs. All data generated or analysed during this study are included in this published article Multiple regulatory features were considered including protein binding, (and its supplementary information files). RNA samples used in this work have been transcription factor binding sites (TFBS), promoters, enhancers, insulators, histone utilized in its entirety and thus are not available. Raw RNAseq (fastq) files used in this modifications, and DNAse hypersensitive regions (DHS). We classified these into work have been deposited in the UCSF Data Share Server [https://doi.org/10.7272/ three broad groups representing promoter/enhancer/transcription (PEX), inert/ Q6HQ3X3M]. quiescent (ZQI), and repressor (R, Supplementary Table 6). We next computed weighted SNP-based scores based on the genotype and number of risk alleles to quantitate the regulatory influence of variation at each SNP. The weights were Code availability counted as positive if there was evidence that the region promotes transcription Computer code used in this study is available upon request. and as negative if there was evidence of repression. The weights were then normalized by the total number of experiments conducted for each respective cell type to remove bias against well-studied cell types. These weighted weights (WW) Received: 7 August 2018 Accepted: 26 March 2019 were summed up across SNPs resulting in a sum of weighted weights (SWW, or predicted regulatory effect -PRE-) per gene per region in each cell type. The WW concept derives from the fact that the sum of the effects of neighboring SNP to a given gene is weighted twice. The first time we weight the number of experiments reported in ENCODE or REP for a given SNP-gene pair (e.g. we assign more value to a relationship that has been reported in 10 independent experiments, to another References that has been reported just once). The second time, we weight the evidence 1. Paul, D. S. et al. Maps of open chromatin guide the functional follow-up of stemming from all SNPs nearby a gene (depending on the LD structure there could genome-wide association signals: application to hematological traits. PLoS be ~100 SNP near a given gene). A gene with a positive score indicates there is Genet 7, e1002139 (2011). evidence that the region containing the MS-associated SNP(s) is actively 2. Hormozdiari, F. et al. Colocalization of GWAS and eQTL signals detects target influencing its transcription in that particular cell type and vice-versa. genes. Am. J. Hum. Genet 99, 1245–1260 (2016). All computations were performed in parallel using the 7400-core QB3 computer cluster at UCSF. 3. Maurano, M. T. et al. Systematic localization of common disease-associated variation in regulatory DNA. Science 337, 1190–1195 (2012). 4. Shooshtari, P., Huang, H. & Cotsapas, C. Integrative genetic and epigenetic Protein interaction network-based pathway analysis (PINBPA). An experi- analysis uncovers regulatory mechanisms of autoimmune disease. Am. J. Hum. mentally determined human protein interactome consisting of 15,783 nodes and Genet 101, 75–86 (2017). 455,321 edges was used for this part of the analysis21. We loaded the network into 5. Cantor, R. M., Lange, K. & Sinsheimer, J. S. Prioritizing GWAS results: a Cytoscape48 and created cell-specific sub-networks using gene expression values review of statistical methods and recommendations for their application. from elsewhere. Specifically, we filtered interactions realized only by gene products Am. J. Hum. Genet 86, 6–22 (2010). expressed in a given cell type, by using RNAseq expression profiles from Kitsak 6. Wang, K., Li, M. & Hakonarson, H. Analysing biological pathways in genome- et al.49. Thus, for the T cell interactome, we only retrieved interactions between wide association studies. Nat. Rev. Genet 11, 843–854 (2010). proteins known to be expressed by any T cell subset present in Kitsak et al. In the case of CNS, while gene expression data is sufficiently granular (profiles for dif- 7. Huang, H., Chanda, P., Alonso, A., Bader, J. S. & Arking, D. E. Gene-based ferent brain cell types and regions exist), epigenomic data for CNS cells/tissues in tests of association. PLoS Genet 7, e1002177 (2011). ENCODE or REP is very sparse, thus we decided to merge all data into a single 8. Carbonetto, P. & Stephens, M. Integrated enrichment analysis of variants and CNS category. pathways in genome-wide association studies indicates central role for IL-2 Next, we loaded the gene-level PRE for each cell type as node attributes and signaling genes in type 1 diabetes, and cytokine signaling genes in Crohn’s conducted a topological analysis by selecting the subnetwork corresponding to the disease. PLoS Genet 9, e1003770 (2013). largest connected component of nodes with positive PRE (those with negative 9. Li, M. X., Kwan, J. S. & Sham, P. C. HYST: a hybrid set-based test for genome- scores are assumed not to be expressed, and thus not to be active players of the wide association studies, with application to protein-protein interaction-based interactome). To eliminate noise from very small or loosely unconnected networks, association analysis. Am. J. Hum. Genet 91, 478–488 (2012). only those with more than 15 nodes were considered. Sensitivity analysis was 10. Chen, M., Cho, J. & Zhao, H. Incorporating biological pathways via a Markov performed by defining different thresholds on PRE values (10th, 25th, 50th random field model in genome-wide association studies. PLoS Genet 7, percentiles) and building networks with only proteins exceeding these thresholds e1001353 (2011). (Supplementary Fig S1). Individual network analysis was performed considering 11. International Multiple Sclerosis Genetics, Consortim. Network-based multiple differing sets of the potentially associated SNPs identified in our recently completed sclerosis pathway analysis with GWAS data from 15,000 cases and 30,000 meta-analysis; those SNP that showed statistically significant evidence of controls. Am. J. Hum. Genet 92, 854–865 (2013). NATURE COMMUNICATIONS | (2019)10:2236 | https://doi.org/10.1038/s41467-019-09773-y | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09773-y 12. Wang, L., Matsushita, T., Madireddy, L., Mousavi, P. & Baranzini, S. E. 41. Agerbo, E. et al. Polygenic risk score, parental socioeconomic status, family PINBPA: cytoscape app for network analysis of GWAS data. Bioinformatics history of psychiatric disorders, and the risk for schizophrenia: a Danish 31, 262–264 (2015). population-based study and meta-analysis. JAMA Psychiatry 72, 635–641 13. Jannot, A. S., Ehret, G. & Perneger, T. P < 5 × 10(-8) has emerged as a (2015). standard of statistical significance for genome-wide association studies. J. Clin. 42. Radice, P., Pharoah, P. D. & Peterlongo, P. Personalized testing based on Epidemiol. 68, 460–465 (2015). polygenic risk score is promising for more efficient population-based 14. Panagiotou, O. A. & Ioannidis, J. P. Genome-Wide Significance Project What screening programs for common oncological diseases. Ann. Oncol. 27, should the genome-wide significance threshold be? Empirical replication of 369–370 (2016). borderline genetic associations. Int J. Epidemiol. 41, 273–286 (2012). 43. Shieh, Y. et al. Breast cancer risk prediction using a clinical risk model and 15. ENCODE Project Consortium The ENCODE (ENCyclopedia Of DNA polygenic risk score. Breast Cancer Res Treat. 159, 513–525 (2016). Elements) Project. Science 306, 636–640 (2004). 44. Natarajan, P. et al. Polygenic risk score identifies subgroup with higher burden 16. Roadmap Epigenomics, C. et al. Integrative analysis of 111 reference human of atherosclerosis and greater relative benefit from statin therapy in the epigenomes. Nature 518, 317–330 (2015). primary prevention setting. Circulation 135, 2091–2101 (2017). 17. Patsopoulos, N. et al. The multiple sclerosis genomic map: role of peripheral 45. Fernandez-Ruiz, I. Coronary artery disease: new polygenic risk score improves immune cells and resident microglia in susceptibility. Preprint at http://www. prediction of CHD. Nat. Rev. Cardiol. 13, 697 (2016). biorxiv.org/content/biorxiv/early/2017/07/13/143933.full.pdf (2017). 46. Boyle, A. P. et al. Annotation of functional variation in personal genomes 18. Steinberg, M. W., Cheung, T. C. & Ware, C. F. The signaling networks of the using RegulomeDB. Genome Res 22, 1790–1797 (2012). herpesvirus entry mediator (TNFRSF14) in immune regulation. Immunol. 47. De Jager, P. L. et al. ImmVar project: insights and design considerations for Rev. 244, 169–187 (2011). future studies of “healthy” immune variation. Semin Immunol. 27, 51–57 19. Grewal, I. S. & Flavell, R. A. CD40 and CD154 in cell-mediated immunity. (2015). Annu Rev. Immunol. 16, 111–135 (1998). 48. Shannon, P. et al. Cytoscape: a software environment for integrated models 20. Baranzini, S. E. et al. Pathway and network-based analysis of genome-wide of biomolecular interaction networks. Genome Res. 13, 2498–2504 (2003). association studies in multiple sclerosis. Hum. Mol. Genet 18, 2078–2090 49. Kitsak, M. et al. Tissue specificity of human disease module. Sci. Rep. 6, 35241 (2009). (2016). 21. Lee, I., Blom, U. M., Wang, P. I., Shim, J. E. & Marcotte, E. M. Prioritizing 50. Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, candidate disease genes by network-based boosting of genome-wide 15–21 (2013). association data. Genome Res. 21, 1109–1121 (2011). 51. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and 22. Rossin, E. J. et al. Proteins encoded in genomic regions associated with dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014). immune-mediated disease physically interact and suggest underlying biology. 52. Liao, Y., Smyth, G. K. & Shi, W. featureCounts: an efficient general purpose PLoS Genet. 7, e1001273 (2011). program for assigning sequence reads to genomic features. Bioinformatics 30, 23. Boyle, E. A., Li, Y. I. & Pritchard, J. K. An expanded view of complex traits: 923–930 (2014). from polygenic to omnigenic. Cell 169, 1177–1186 (2017). 24. Lenschow, D. J., Walunas, T. L. & Bluestone, J. A. CD28/B7 system of T cell costimulation. Annu Rev. Immunol. 14, 233–258 (1996). 25. Stevenson, C. et al. Essential role of Elmo1 in Dock2-dependent lymphocyte Acknowledgements migration. J. Immunol. 192, 6062–6070 (2014). This investigation was supported in part by the following sources: NIH/NINDS awards 26. Anwar, A. et al. Mer tyrosine kinase (MerTK) promotes macrophage survival R01NS088155 and 1R01NS099240, the Valhalla Charitable Foundation, and the Heidrich following exposure to oxidative stress. J. Leukoc. Biol. 86, 73–79 (2009). Family and Friends Foundation (Sergio E. Baranzini). US National Multiple Sclerosis 27. Stienne, C. et al. Foxo3 transcription factor drives pathogenic T helper 1 Society (TA 3056-A-2), the Harvard NeuroDiscovery Center and an Intel Parallel differentiation by inducing the expression of Eomes. Immunity 45, 774–787 Computing Center award (Nikolaos A. Patsopoulos). Swedish Medical Research Council; (2016). Swedish Research Council for Health, Working Life and Welfare, Knut and Alice 28. Lupar, E. et al. Eomesodermin expression in CD4+ T cells restricts peripheral Wallenberg Foundation, AFA insurance, Swedish Brain Foundation, the Swedish Asso- Foxp3 induction. J. Immunol. 195, 4742–4752 (2015). ciation for Persons with Neurological Disabilities. Cambridge NIHR Biomedical Research 29. Endo, Y. et al. Eomesodermin controls interleukin-5 production in memory T Centre, UK Medical Research Council (G1100125) and the UK MS society (861/07). helper 2 cells through inhibition of activity of the transcription factor GATA3. NIH/NINDS: R01 NS049477, NIH/NIAID: R01 AI059829, NIH/NIEHS: R01 ES0495103. Immunity 35, 733–745 (2011). Research Council of Norway grant 196776 and 240102. NINDS/NIH R01NS088155. Oslo 30. Parnell, G. P. et al. The autoimmune disease-associated transcription factors MS association. Research Council KU Leuven, Research Foundation Flanders. AFM, EOMES and TBX21 are dysregulated in multiple sclerosis and define a AFM-Généthon, CIC, ARSEP, ANR-10-INBS-01 and ANR-10-IAIHU-06. Research molecular subtype of disease. Clin. Immunol. 151, 16–24 (2014). Council KU Leuven, Research Foundation Flanders. Inserm ATIP-Avenir Fellowship 31. Smets, I. et al. Multiple sclerosis risk variants alter expression of and Connect-Talents Award. German Ministry for Education and Research, German co-stimulatory genes in B cells. Brain 141, 786–796 (2018). Competence Network MS (BMBF KKNMS). Oslo MS association, Research Council of 32. Webb, L. M., Foxwell, B. M. & Feldmann, M. Interleukin-7 activates human Norway grant 196776 and 240102. Dutch MS Research Foundation. TwinsUK is funded naive CD4+ cells and primes for interleukin-4 production. Eur. J. Immunol. by the Wellcome Trust, Medical Research Council, European Union, the National 27, 633–640 (1997). Institute for Health Research (NIHR)-funded BioResource, Clinical Research Facility and 33. Guo, L. et al. Innate immunological function of TH2 cells in vivo. Nat. Biomedical Research Centre based at Guy’s and St Thomas’ NHS Foundation Trust in Immunol. 16, 1051–1059 (2015). partnership with King’s College London. German Ministry for Education and Research, 34. Davis, R. S. Fc receptor-like molecules. Annu Rev. Immunol. 25, 525–560 German Competence Network MS (BMBF KKNMS). Italian Foundation of Multiple (2007). Sclerosis (FISM). NMSS (RG 4680A1/1). German Ministry for Education and Research, 35. De Jager, P. L. et al. Integration of genetic risk factors into a clinical algorithm German Competence Network MS (BMBF KKNMS). Lundbeck Foundation and Benzon for multiple sclerosis susceptibility: a weighted genetic risk score. Lancet Foundation. Neurol. 8, 1111–1119 (2009). 36. Gourraud, P. A. et al. Aggregation of multiple sclerosis genetic risk variants in multiple and single case families. Ann. Neurol. 69, 65–74 (2011). Author contribution 37. Pers, T. H. et al. Biological interpretation of genome-wide association studies L.F.B., L.B., F.M-B., D.B., M.C., G.C., B.A.C.C., S.D., K.D., P.D., D.E., F.E., B.F., A.G., using predicted gene functions. Nat. Commun. 6, 5890 (2015). B.D., G.H., C.H., B.H., R.H., D.H., N.I., S.K., J.K., M.K., I.K., R.M., A.O., A.P., M.P-V., 38. Watanabe, K., Taskesen, E., van Bochoven, A. & Posthuma, D. Functional J.S., T.O., B.V.T., G.S., H.F.H., A.C., S.L.H., D.A.H., F.Z., P.D., S.S., and J.R.O. directly mapping and annotation of genetic associations with FUMA. Nat. Commun. contributed to sample acquisition. S.B., J.M., K.K., S.J.C., and S.E.B. performed DNA/ 8, 1826 (2017). RNA processing. A.B., J.M., S.J.C., P.D., and S.E.B. performed genotyping and/or 39. Lamparter, D., Marbach, D., Rueedi, R., Kutalik, Z. & Bergmann, S. Fast and RNAseq. L.M., N.A.P., C.C., S.D.B., A.B., J.M., K.K., X.J., T.F.A., T.B., F.M-B., D.B., F.B., rigorous computation of gene and pathway scores from SNP-based summary E.G.C., B.A.C.C., C.G., A.G., P-A.G., J.H., N.I., C.M.L., M.L., F.L., R.H., J.S., H.F.H., S. statistics. PLoS Comput Biol. 12, e1004714 (2016). L.H., D.A.H., F.Z., P.D., S.S., J.R.O., and S.E.B. performed analysis. L.M., N.A.P., C.C., 40. Amare, A. T. International Consortium on Lithium Genetics (ConLi+Gen) A.B., J.M., K.K., A.S., S.J.C., J.H., and S.E.B. we involved in data handling (clinical, Association of polygenic score for schizophrenia and hla antigen and demographic or genotypic). N.P., C.C., P-A.G., J.H., B.H., I.K., R.H., A.I., B.V.T., G.S., inflammation genes with response to lithium in bipolar affective disorder: A H.F.H., A.C., S.L.H., D.A.H., F.Z., P.D., S.S., J.R.O., and S.E.B. designed the study. L.M., Genome-Wide Association Study. JAMA Psychiatry 75, 65–74 (2017). C.C., S.D.B., J.M., K.K., S.L.H., S.S., J.R.O., and S.E.B. wrote the paper. 10 NATURE COMMUNICATIONS | (2019)10:2236 | https://doi.org/10.1038/s41467-019-09773-y | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09773-y ARTICLE Additional information Open Access This article is licensed under a Creative Commons Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- Attribution 4.0 International License, which permits use, sharing, 019-09773-y. adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Competing interests: The authors declare no competing interests. Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless Reprints and permission information is available online at http://npg.nature.com/ indicated otherwise in a credit line to the material. If material is not included in the reprintsandpermissions/ article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from Journal peer review information: Nature Communications thanks the anonymous the copyright holder. To view a copy of this license, visit http://creativecommons.org/ reviewers for their contribution to the peer review of this work. Peer reviewer reports are licenses/by/4.0/. available. Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in © The Author(s) 2019 published maps and institutional affiliations. International Multiple Sclerosis Genetics Consortium Lohith Madireddy1, Nikolaos A. Patsopoulos2,3, Chris Cotsapas4,5, Steffan D. Bos6,7, Ashley Beecham8, Jacob McCauley8,9, Kicheol Kim1, Xiaoming Jia1, Adam Santaniello1, Stacy J. Caillier1, Till F.M. Andlauer10,11, Lisa F. Barcellos12, Tone Berge7,13, Luisa Bernardinelli14, Filippo Martinelli-Boneschi15,16, David R. Booth17, Farren Briggs18, Elisabeth G. Celius7,19, Manuel Comabella20, Giancarlo Comi21, Bruce A.C. Cree1, Sandra D’Alfonso22, Katrina Dedham23, Pierre Duquette24, Efthimios Dardiotis25, Federica Esposito21, Bertrand Fontaine26,27, Christiane Gasperi10, An Goris28, Bénédicte Dubois28, Pierre-Antoine Gourraud29,30, Georgios Hadjigeorgiou31, Jonathan Haines18, Clive Hawkins32,37, Bernhard Hemmer10, Rogier Hintzen33,34, Dana Horakova35, Noriko Isobe36, Seema Kalra32,37, Jun-ichi Kira36, Michael Khalil38, Ingrid Kockum39, Christina M. Lill40,41, Matthew R. Lincoln4, Felix Luessi40, Roland Martin42, Annette Oturai43, Aarno Palotie44,45, Margaret A. Pericak-Vance8,9, Roland Henry1, Janna Saarela44, Adrian Ivinson46, Tomas Olsson39, Bruce V. Taylor47, Graeme J. Stewart17, Hanne F. Harbo6,7, Alastair Compston48, Stephen L. Hauser1, David A. Hafler4, Frauke Zipp40, Philip De Jager3,49, Stephen Sawcer48, Jorge R. Oksenberg1 & Sergio E. Baranzini 1,50 1 Weill Institute for Neurosciences, Department of Neurology, University of California San Francisco, San Francisco, CA 94158, USA. 2Systems Biology and Computer Science Program, Ann Romney Center for Neurological Diseases, Department of Neurology, and Division of Genetics, Department of Medicine, Brigham & Women’s Hospital, Harvard Medical School, Boston, MA 02115, USA. 3Broad Institute of Harvard University and Massachusetts Institute of Technology, Cambridge, MA 02142, USA. 4Departments of Neurology, Yale School of Medicine, 300 George St, New Haven, CT 06511, USA. 5Department of Genetics, Yale School of Medicine, 300 George St, New Haven, CT 06511, USA. 6Institute of Clinical Medicine, University of Oslo, Oslo 0318, Norway. 7Department of Neurology, Oslo University Hospital, Oslo 0424, Norway. 8John P. Hussman Institute for Human Genomics, University of Miami, Miller School of Medicine, Miami, FL 33136, USA. 9Dr. John T. Macdonald Foundation Department of Human Genetics, University of Miami, Miller School of Medicine, Miami, FL 33136, USA. 10Department of Neurology, Klinikum rechts der Isar, School of Medicine, Technical University of Munich, 81675 Munich, Germany. 11Munich Cluster for Systems Neurology (SyNergy), 81377 Munich, Germany. 12Division of Epidemiology, School of Public HealthUniversity of California, 324 Stanley Hall, MC#3220, Berkeley, CA 94720, USA. 13Department of Mechanical, Electronics and Chemical Engineering, Oslo Metropolitan University, Oslo 0167, Norway. 14Section of Biostatistics, Neurophyisiology and Psychiatry, Unit of medical and genomic statistics, Universita of Pavia, Pavia 27100, Italy. 15Department of Biomedical Sciences for Health, University of Milan, Milan 20133, Italy. 16MS Research Unit and Department of Neurology, IRCCS Policlinico San Donato, San Donato Milanese, Milan 20097, Italy. 17Faculty of Medicine, Westmead Clinical School, The Westmead Institute for Medical Research, Sydney, NSW 2145, Australia. 18Department of Quantitative and Population Health Sciences, School of Medicine, Case Western Reserve University, Cleveland, OH 44106, USA. 19Institute of Health and Society, University of Oslo, Oslo 0318, Norway. 20Servei de Neurologia-Neuroimmunologia. Centre d’Esclerosi Múltiple de Catalunya (Cemcat), Institut de Recerca Vall d’Hebron (VHIR), Hospital Universitari Vall d’Hebron, Universitat Autònoma de Barcelona, Barcelona 08035, Spain. 21Department of Neurology, San Raffaele Scientific Institute, Milan 20132, Italy. 22Department of Health Sciences, UPO University, Novara 28100, Italy. 23Department of Clinical Neurosciences. Neurology Unit, University of Cambridge, Cambridge CB2 1QW, UK. 24Faculté de médecine, MS Clinic Centre Hospitalier de l’, Université de Montréal. Université de Montréal Montreal, Montreal QC H3A 1G1, Canada. 25Department of Neurology, Laboratory of Neurogenetics, University of Thessaly, University Hospital of Larissa, Larissa 41223, Greece. 26Department of Neurology, University Hopital Pitié-Salpêtrière, Paris 75013, France. 27UMR 1127, Sorbonne-Université, INSERM, University Hopital Pitié-Salpêtrière, Paris 75013, France. 28KU Leuven Department of Neurosciences, Laboratory for Neuroimmunology, Leuven 3000, Belgium. 29Université de Nantes, INSERM, Centre de Recherche en Transplantation et Immunologie, UMR 1064, ATIP-Avenir, Equipe 5, Nantes F-44093, France. 30CHU de Nantes, INSERM, CIC 1413, Pôle Hospitalo-Universitaire 11: Santé Publique, Clinique des données, Nantes F-44093, France. 31Department of Neurology, Medical School, University of Cyprus, Nicosia 587G+X2, Cyprus. 32Institute for Science & Technology NATURE COMMUNICATIONS | (2019)10:2236 | https://doi.org/10.1038/s41467-019-09773-y | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09773-y in Medicine, Keele University, Keele ST5 5GB, UK. 33Department of Neurology, Erasmus MC Dr Molewaterplein 40, Rotterdam 3015 GD, The Netherlands. 34Department of Immunology, Erasmus MC, Rotterdam 3015 GD, The Netherlands. 35First Faculty of Medicine, Department of Neurology and Center of Clinical Neuroscience, Charles University and General University Hospital, Prague 3CFG+RJ, Czech Republic. 36Department of Neurology, Kyushu University, Kyushu 812-0053, Japan. 37Royal Stoke MS Centre of Excellence, University Hospital North Midlands, Stoke-on-Trent, ST4 6QG, UK. 38Department of Neurology, Medical University of Graz, Graz A-8036, Austria. 39Department of Clinical Neuroscience and Center for Molecular Medicine, Karolinska Institutet, Stockholm 17176, Sweden. 40Department of Neurology, University Medical Center of the Johannes Gutenberg University Mainz, Mainz 55131, Germany. 41Genetic and Molecular Epidemiology Group, Lübeck Interdisciplinary Platform for Genome Analytics, Institutes of Neurogenetics and Cardiogenetics, University of Lübeck, Lübeck 23562, Germany. 42Neuroimmunology and MS Research (nims), Department of Neurology, University Zurich, Zürich 8006, Switzerland. 43Department of Neurology, section 2082, Rigshospitalet, Danish Multiple Sclerosis Center, University of Copenhagen, Copenhagen 2100, Denmark. 44Institute for Molecular Medicine Finland (FIMM), University of Helsinki, Helsinki FIN-00014, Finland. 45Harvard University, Center for Human Genetic Research, Boston, MA 02115, USA. 46UK Dementia Research Institute at University College London, London WC1E6BT, UK. 47Menzies Institute for Medical Research, University of Tasmania, Hobart, TAS 7000, Australia. 48Department of Clinical Neurosciences, Cambridge Biomedical Campus, University of Cambridge, Cambridge CB2 0QQ, UK. 49Department of Neurology, Center for Translational and Computational Neuroimmunology and Multiple Sclerosis Center, Columbia University Medical Center, New York, NY 10032, USA. 50Bakar Institute for Computational Health Science. University of California San Francisco, San Francisco, CA 94158, USA 12 NATURE COMMUNICATIONS | (2019)10:2236 | https://doi.org/10.1038/s41467-019-09773-y | www.nature.com/naturecommunications