diff --git a/main_text/5_AD_xQTL_genes_cis_trans/staging/APOE_code/APOE_cs_neuropatho.R b/main_text/5_AD_xQTL_genes_cis_trans/staging/APOE_code/APOE_cs_neuropatho.R new file mode 100644 index 00000000..f6bf1af5 --- /dev/null +++ b/main_text/5_AD_xQTL_genes_cis_trans/staging/APOE_code/APOE_cs_neuropatho.R @@ -0,0 +1,263 @@ +source('../../utils/r_utils.R') +out<-'main_text/5_AD_xQTL_genes_cis_trans/staging/APOE/' +dir.create(out) +#APOE4 independant APOE cs are associated with neuropatho in ROSMAP ? + +res_indep_cs<-readRDS('main_text/5_AD_xQTL_genes_cis_trans/staging/APOE/xqtl_only_APOE_all_cohorts_independent_sets_after_imputation_no_sQTL_only.rds') +length(res_indep_cs$variant_id) +indep_cs<-rbindlist(lapply(names(res_indep_cs$variant_id),function(n)data.table(cs_name=n,variant_id=res_indep_cs$variant_id[[n]], + gwas_source=res_indep_cs$ad_gwas[[which(names(res_indep_cs$variant_id)==n)]], + molecular_outcome=res_indep_cs$outcome[[n]]))) +indep_cs[,cs_id:=paste0('set_',str_extract(cs_name,'[0-9]+$'))] +indep_cs[,chr:=seqid(variant_id)] +indep_cs[,pos:=pos(variant_id)] +indep_cs[,variant_id2:=str_replace_all(variant_id,':','_')|>str_replace('_',':')] + +#extract genotype for those CS +fwrite(indep_cs[order(chr,pos)][,.(variant_id2)],fp(out,'apoe4_indep_snps.tsv'),sep = '\t',col.names=FALSE) + +plink_file<-'/projectnb/tcwlab-adsp/member/adpelle1/projects/fungen-xqtl/ref-data/ROSMAP/ROSMAP_NIA_geno/ROSMAP_NIA_WGS.leftnorm.bcftools_qc.plink_qc' + +cmd=paste('plink --bfile',tools::file_path_sans_ext(plink_file), + '--extract',fp(out,'apoe4_indep_snps.tsv'), + '--recode vcf', + '--out',fp(out,'ROSMAP_geno_apoe4_indep_snps')) + +system(cmd) + +#format the genotype +donors_apoecs<-fread(fp(out,'ROSMAP_geno_apoe4_indep_snps.vcf')) +dim(donors_apoecs)#216 1162 + +donors_apoecs<-melt(donors_apoecs[,-c(6:9)], + id.vars = c('#CHROM','POS','ID','REF','ALT'), + variable.name = 'IID',value.name = 'geno') +donors_apoecs[,dose:=str_count(geno,'1')] + +donors_apoecs[,variant_id2:=ID] +donors_apoecs<-merge(donors_apoecs,indep_cs,by='variant_id2') + + +#add the clinical covariates +clin<-fread('/projectnb/tcwlab/ShareSpace/ROSMAP_Neuropatho/ROSMAP_clinical.csv') +bios<-fread('/projectnb/tcwlab/ShareSpace/ROSMAP_Neuropatho/ROSMAP_biospecimen_metadata.csv') #for WGS samples ID - IndividualID link +clinw<-merge(clin,bios[assay=='wholeGenomeSeq'][,.(individualID,specimenID)],by='individualID') +complete<-fread('/projectnb/tcwlab/ShareSpace/ROSMAP_Neuropatho/ROSMAP_xqtl_complete_samples_covariates_sex_death_pmi_study.csv') +setnames(complete,'sample_id','specimenID') +clinw<-merge(clinw[,-c('age_death','pmi','msex')],complete,by='specimenID') + +donors_apoecs[,specimenID:=str_remove(IID,'0_')] + +donors_apoecs<-merge(donors_apoecs,clinw,by='specimenID') + +unique(donors_apoecs$individualID)|>length()#1116 + +#add neuropatho covs +#we test only a subset of the covariates +neuropath<-fread('/projectnb/tcwlab/ShareSpace/ROSMAP_Neuropatho/dataset_707_basic_02-10-2022.n3711.clean.plink.n2409.28FEB2022.txt') +dim(neuropath) +neuropathf<-neuropath[,.SD,.SDcols = c('projid','amyloid_sqrt','tangles_sqrt','cogng_path_slope','plaq_d','plaq_n_sqrt','nft_sqrt','tdp_cs_6reg','tdp_dn_6reg','tdp_st4','caa_4gp','caa_neo4')] + +#add those phenotype to the genotypes +donors_apoecs<-merge(donors_apoecs,neuropathf,all.x = TRUE,by='projid') + +#neuropatho analysis correcting for APOE4 /2 gentoyep +donors_apoecs[,apoe4_dose:=str_count(apoe_genotype,'4')] +donors_apoecs[,apoe2_dose:=str_count(apoe_genotype,'2')] + +#prep/reformat the phenotypes +donors_apoecs[,cogdx_binary:=ifelse(cogdx%in%c(4,5),'AD',ifelse(cogdx%in%1:3,'No AD',NA))] #binarize the cognitive status + +donors_apoecs[,cogdx_merge:=ifelse(cogdx%in%c(4,5),'AD',ifelse(cogdx%in%2:3,'MCI',ifelse(cogdx==1,'NL',NA)))] # three level cognitive status +donors_apoecs[,cogdx_merge:=factor(cogdx_merge,levels = c('NL','MCI','AD'))] +donors_apoecs[cogdx!=6,cogdx_mergenum:=ifelse(cogdx_merge=='MCI',2,ifelse(cogdx_merge=='AD',3,1))] + +donors_apoecs[,age_first_ad_dx_num:=ifelse(age_first_ad_dx=='90+',91,as.numeric(age_first_ad_dx))] +donors_apoecs[,other_dementia:=cogdx==6] +donors_apoecs[,other_CI:=cogdx%in%c(3,5)] +donors_apoecs[cogdx%in%c(1,4,5),ADvsNL:=cogdx%in%c(4,5)] +donors_apoecs[cogdx%in%c(1,2,3),MCIvsNL:=cogdx%in%c(2,3)] + +donors_apoecs[,ceradsc_pos:=-ceradsc] +donors_apoecs[,braaksc06:=ifelse(braaksc==0,1,ifelse(braaksc==6,5,braaksc))] + +donors_apoecs[,age_first_ad_dx_num:=ifelse(age_first_ad_dx=='90+',91,as.numeric(age_first_ad_dx))] +donors_apoecs[(cogdx_binary=='AD'),age_first_clinad_dx:=age_first_ad_dx_num] +donors_apoecs[(cogdx_binary=='AD'),cts_mmse30_lv_clinad:=cts_mmse30_lv] + +donors_apoecs[,age_at_visit_num:=ifelse(age_at_visit_max=='90+', + 91,as.numeric(age_at_visit_max))] +donors_apoecs[dcfdx_lv!=6,dcfdx_lv_no6:=dcfdx_lv] + +donors_apoecs[,apoe2_r:=cor(dose,apoe2_dose,use = 'pairwise.complete.obs'),by='ID'] +donors_apoecs[,apoe4_r:=cor(dose,apoe4_dose,use = 'pairwise.complete.obs'),by='ID'] + +fwrite(donors_apoecs,fp(out,'ROSMAP_geno_apoecs_adpathology.csv.gz')) + + +#outcome to test with Lm +covs_lin<-c('ceradsc_pos','braaksc06', + 'dcfdx_lv_no6','cts_mmse30_lv','age_death', + 'cogdx_mergenum','apoe4_dose','apoe2_dose', + 'cts_mmse30_lv_clinad','age_first_ad_dx_num', + 'age_first_clinad_dx','amyloid_sqrt','tangles_sqrt', + 'cogng_path_slope', + 'plaq_d','plaq_n_sqrt','nft_sqrt','tdp_cs_6reg','tdp_dn_6reg','tdp_st4','caa_4gp','caa_neo4' +) + +#outcome to test with logistic regression +covs_logis<-c('MCIvsNL','ADvsNL') + +#outcome that are measured before death +atlastvisit=c('cts_mmse30_lv','cts_mmse30_lv_clinad')#outcome that need to be adjusted for age at last visit + +beforedeath<-c('age_death','cts_mmse30_lv','cts_mmse30_lv_clinad','age_first_ad_dx_num','age_first_clinad_dx','cogng_path_slope') + +#Adjusting for both APOE4 and APOE2#### +apoecs_clin<-rbindlist(lapply(covs_lin,function(outc){ + message(outc) + if(outc%in%atlastvisit){ + donors_apoecs[,summary(lm(unlist(.SD)~dose+msex+educ+age_at_visit_num+apoe4_dose+apoe2_dose))$coefficients|>data.table(keep.rownames = 'cov'), + by='variant_id2',.SDcols=outc][cov=='dose'][,outcome:=outc] + + }else if(outc%in%beforedeath){ + donors_apoecs[,summary(lm(unlist(.SD)~dose+msex+educ+apoe4_dose+apoe2_dose))$coefficients|>data.table(keep.rownames = 'cov'), + by='variant_id2',.SDcols=outc][cov=='dose'][,outcome:=outc] + }else{ + + donors_apoecs[,summary(lm(unlist(.SD)~dose+msex+educ+age_death+apoe4_dose+apoe2_dose))$coefficients|>data.table(keep.rownames = 'cov'), + by='variant_id2',.SDcols=outc][cov=='dose'][,outcome:=outc] + } + +})) +apoecs_clin[,zscore:=`t value`] +apoecs_clin[,pval:=`Pr(>|t|)`] + +apoecs_clogit<-rbindlist(lapply(covs_logis,function(outc){ + message(outc) + donors_apoecs[,summary(glm(unlist(.SD)~dose+msex+educ+age_death+apoe4_dose+apoe2_dose,family = binomial))$coefficients|>data.table(keep.rownames = 'cov'), + by='variant_id2',.SDcols=outc][cov=='dose'][,outcome:=outc] + +})) +apoecs_clogit[,zscore:=`z value`] +apoecs_clogit[,pval:=`Pr(>|z|)`] + +apoecs_clin<-rbind(apoecs_clin,apoecs_clogit,fill=TRUE) + +apoecs_clin<-merge(apoecs_clin,indep_cs,by='variant_id2') +apoecs_clin[,padj:=p.adjust(pval,n = length(unique(indep_cs$cs_name))),by=c('outcome','variant_id2')] +apoecs_clin[padj<0.05] + + +apoecs_clin[!outcome%in%c('apoe4_dose','apoe2_dose')][padj<0.05] + +apoecs_clin[padj<0.05][!outcome%in%c('apoe4_dose','apoe2_dose')]$cs_id |>unique()|>length()#3/19 + +fwrite(apoecs_clin,fp(out,'res_lm_apoecs_adpathology_apoe4_2_correction.csv.gz')) + + +unique(apoecs_clin[padj<0.05][!outcome%in%c('apoe4_dose','apoe2_dose')][order(cs_id,outcome,padj)][,.(cs_id,outcome,padj,gwas_source,molecular_outcome)],by=c('cs_id','outcome')) + +# cs_id outcome padj gwas_source +# +# 1: set_27 caa_4gp 0.004708944 AD_Kunkle_Stage1_2019; AD_Wightman_Full_2021; AD_Wightman_Excluding23andMe_2021; AD_Wightman_ExcludingUKBand23andME_2021 +# 2: set_27 caa_neo4 0.007830511 AD_Kunkle_Stage1_2019; AD_Wightman_Full_2021; AD_Wightman_Excluding23andMe_2021; AD_Wightman_ExcludingUKBand23andME_2021 +# 3: set_27 tdp_st4 0.042915291 AD_Kunkle_Stage1_2019; AD_Wightman_Full_2021; AD_Wightman_Excluding23andMe_2021; AD_Wightman_ExcludingUKBand23andME_2021 +# 4: set_31 age_first_ad_dx_num 0.020960086 AD_Bellenguez +# 5: set_57 tangles_sqrt 0.003434581 AD_Bellenguez; AD_Kunkle_Stage1_2019; AD_Wightman_Full_2021; AD_Wightman_Excluding23andMe_2021 +# molecular_outcome +# +# 1: AC_DeJager_eQTL_ENSG00000224916; AC_DeJager_eQTL_ENSG00000234906; BM_10_MSBB_eQTL_ENSG00000104853; Metabrain_Basalganglia_chr19_41840000_47960000_ENSG00000234906; Metabrain_Cerebellum_chr19_41840000_47960000_ENSG00000234906; Metabrain_Cortex_chr19_41840000_47960000_ENSG00000104853; Metabrain_Cortex_chr19_41840000_47960000_ENSG00000234906; Mic_DeJager_eQTL_ENSG00000130208; Mic_DeJager_eQTL_ENSG00000130203; Mic_mega_eQTL_ENSG00000130203; STARNET_eQTL_Mac_ENSG00000130208; STARNET_eQTL_Mac_ENSG00000130203; STARNET_eQTL_Mac_ENSG00000267467; STARNET_eQTL_Mac_ENSG00000224916; STARNET_eQTL_Mac_ENSG00000234906; DeJager_Mic_ENSG00000130203; Kellis_Mic_ENSG00000130203; DeJager_Mic_ENSG00000130208; Kellis_Mic_ENSG00000130208 +# 2: AC_DeJager_eQTL_ENSG00000224916; AC_DeJager_eQTL_ENSG00000234906; BM_10_MSBB_eQTL_ENSG00000104853; Metabrain_Basalganglia_chr19_41840000_47960000_ENSG00000234906; Metabrain_Cerebellum_chr19_41840000_47960000_ENSG00000234906; Metabrain_Cortex_chr19_41840000_47960000_ENSG00000104853; Metabrain_Cortex_chr19_41840000_47960000_ENSG00000234906; Mic_DeJager_eQTL_ENSG00000130208; Mic_DeJager_eQTL_ENSG00000130203; Mic_mega_eQTL_ENSG00000130203; STARNET_eQTL_Mac_ENSG00000130208; STARNET_eQTL_Mac_ENSG00000130203; STARNET_eQTL_Mac_ENSG00000267467; STARNET_eQTL_Mac_ENSG00000224916; STARNET_eQTL_Mac_ENSG00000234906; DeJager_Mic_ENSG00000130203; Kellis_Mic_ENSG00000130203; DeJager_Mic_ENSG00000130208; Kellis_Mic_ENSG00000130208 +# 3: AC_DeJager_eQTL_ENSG00000224916; AC_DeJager_eQTL_ENSG00000234906; BM_10_MSBB_eQTL_ENSG00000104853; Metabrain_Basalganglia_chr19_41840000_47960000_ENSG00000234906; Metabrain_Cerebellum_chr19_41840000_47960000_ENSG00000234906; Metabrain_Cortex_chr19_41840000_47960000_ENSG00000104853; Metabrain_Cortex_chr19_41840000_47960000_ENSG00000234906; Mic_DeJager_eQTL_ENSG00000130208; Mic_DeJager_eQTL_ENSG00000130203; Mic_mega_eQTL_ENSG00000130203; STARNET_eQTL_Mac_ENSG00000130208; STARNET_eQTL_Mac_ENSG00000130203; STARNET_eQTL_Mac_ENSG00000267467; STARNET_eQTL_Mac_ENSG00000224916; STARNET_eQTL_Mac_ENSG00000234906; DeJager_Mic_ENSG00000130203; Kellis_Mic_ENSG00000130203; DeJager_Mic_ENSG00000130208; Kellis_Mic_ENSG00000130208 +# 4: AC_DeJager_eQTL_ENSG00000189114; DLPFC_DeJager_eQTL_ENSG00000189114; Metabrain_Cortex_chr19_41840000_47960000_ENSG00000189114; PCC_DeJager_eQTL_ENSG00000189114; ROSMAP_AC_ENSG00000189114; ROSMAP_DLPFC_ENSG00000189114; ROSMAP_PCC_ENSG00000189114 +# 5: Ast_10_Kellis_eQTL_ENSG00000160007 + + +#set27 is the caa set of Fujita et al? +apoecs_clin[pos(variant_id2)==44946027]#yes + + + +#adjusting for APOE4 only#### +apoecs_clin2<-rbindlist(lapply(covs_lin,function(outc){ + message(outc) + if(outc%in%atlastvisit){ + donors_apoecs[,summary(lm(unlist(.SD)~dose+msex+educ+age_at_visit_num+apoe4_dose))$coefficients|>data.table(keep.rownames = 'cov'), + by='variant_id2',.SDcols=outc][cov=='dose'][,outcome:=outc] + + }else if(outc%in%beforedeath){ + donors_apoecs[,summary(lm(unlist(.SD)~dose+msex+educ+apoe4_dose))$coefficients|>data.table(keep.rownames = 'cov'), + by='variant_id2',.SDcols=outc][cov=='dose'][,outcome:=outc] + }else{ + + donors_apoecs[,summary(lm(unlist(.SD)~dose+msex+educ+age_death+apoe4_dose))$coefficients|>data.table(keep.rownames = 'cov'), + by='variant_id2',.SDcols=outc][cov=='dose'][,outcome:=outc] + } + +})) +apoecs_clin2[,zscore:=`t value`] +apoecs_clin2[,pval:=`Pr(>|t|)`] + +apoecs_clogit2<-rbindlist(lapply(covs_logis,function(outc){ + message(outc) + donors_apoecs[,summary(glm(unlist(.SD)~dose+msex+educ+age_death+apoe4_dose,family = binomial))$coefficients|>data.table(keep.rownames = 'cov'), + by='variant_id2',.SDcols=outc][cov=='dose'][,outcome:=outc] + +})) +apoecs_clogit2[,zscore:=`z value`] +apoecs_clogit2[,pval:=`Pr(>|z|)`] + +apoecs_clin2<-rbind(apoecs_clin2,apoecs_clogit2,fill=TRUE) + +apoecs_clin2<-merge(apoecs_clin2,indep_cs,by='variant_id2') +apoecs_clin2[,padj:=p.adjust(pval,n = length(unique(indep_cs$cs_name))),by=c('outcome','variant_id2')] +apoecs_clin2[padj<0.05] + + +apoecs_clin2[!outcome%in%c('apoe4_dose','apoe2_dose')][padj<0.05] + +apoecs_clin2[padj<0.05][!outcome%in%c('apoe4_dose','apoe2_dose')]$cs_id |>unique()|>length()#3/19 + +fwrite(apoecs_clin2,fp(out,'res_lm_apoecs_adpathology_apoe4_correction.csv.gz')) + + +unique(apoecs_clin2[padj<0.05][!outcome%in%c('apoe4_dose','apoe2_dose')][order(cs_id,cov,padj)][,.(cs_id,outcome,variant_id2,padj,gwas_source,molecular_outcome)],by=c('cs_id','outcome')) +# cs_id outcome variant_id2 padj +# +# 1: set_26 age_first_ad_dx_num chr19:44918487_G_T 0.046608400 +# 2: set_27 caa_4gp chr19:44954310_T_C 0.002833505 +# 3: set_27 caa_neo4 chr19:44954310_T_C 0.004877779 +# 4: set_27 tdp_st4 chr19:44999110_TAAAA_TAA 0.042966548 +# 5: set_31 age_first_ad_dx_num chr19:45114770_ACCC_AC 0.020474889 +# 6: set_57 tangles_sqrt chr19:44840322_G_A 0.020962988 +# gwas_source +# +# 1: AD_Kunkle_Stage1_2019; AD_Wightman_Full_2021; AD_Wightman_Excluding23andMe_2021 +# 2: AD_Kunkle_Stage1_2019; AD_Wightman_Full_2021; AD_Wightman_Excluding23andMe_2021; AD_Wightman_ExcludingUKBand23andME_2021 +# 3: AD_Kunkle_Stage1_2019; AD_Wightman_Full_2021; AD_Wightman_Excluding23andMe_2021; AD_Wightman_ExcludingUKBand23andME_2021 +# 4: AD_Kunkle_Stage1_2019; AD_Wightman_Full_2021; AD_Wightman_Excluding23andMe_2021; AD_Wightman_ExcludingUKBand23andME_2021 +# 5: AD_Bellenguez +# 6: AD_Bellenguez; AD_Kunkle_Stage1_2019; AD_Wightman_Full_2021; AD_Wightman_Excluding23andMe_2021 +# molecular_outcome +# +# 1: AC_DeJager_eQTL_ENSG00000130208; Metabrain_Cortex_chr19_41840000_47960000_ENSG00000130208; STARNET_eQTL_Mac_ENSG00000130208; ROSMAP_AC_ENSG00000130208; ROSMAP_DLPFC_ENSG00000130208 +# 2: AC_DeJager_eQTL_ENSG00000224916; AC_DeJager_eQTL_ENSG00000234906; BM_10_MSBB_eQTL_ENSG00000104853; Metabrain_Basalganglia_chr19_41840000_47960000_ENSG00000234906; Metabrain_Cerebellum_chr19_41840000_47960000_ENSG00000234906; Metabrain_Cortex_chr19_41840000_47960000_ENSG00000104853; Metabrain_Cortex_chr19_41840000_47960000_ENSG00000234906; Mic_DeJager_eQTL_ENSG00000130208; Mic_DeJager_eQTL_ENSG00000130203; Mic_mega_eQTL_ENSG00000130203; STARNET_eQTL_Mac_ENSG00000130208; STARNET_eQTL_Mac_ENSG00000130203; STARNET_eQTL_Mac_ENSG00000267467; STARNET_eQTL_Mac_ENSG00000224916; STARNET_eQTL_Mac_ENSG00000234906; DeJager_Mic_ENSG00000130203; Kellis_Mic_ENSG00000130203; DeJager_Mic_ENSG00000130208; Kellis_Mic_ENSG00000130208 +# 3: AC_DeJager_eQTL_ENSG00000224916; AC_DeJager_eQTL_ENSG00000234906; BM_10_MSBB_eQTL_ENSG00000104853; Metabrain_Basalganglia_chr19_41840000_47960000_ENSG00000234906; Metabrain_Cerebellum_chr19_41840000_47960000_ENSG00000234906; Metabrain_Cortex_chr19_41840000_47960000_ENSG00000104853; Metabrain_Cortex_chr19_41840000_47960000_ENSG00000234906; Mic_DeJager_eQTL_ENSG00000130208; Mic_DeJager_eQTL_ENSG00000130203; Mic_mega_eQTL_ENSG00000130203; STARNET_eQTL_Mac_ENSG00000130208; STARNET_eQTL_Mac_ENSG00000130203; STARNET_eQTL_Mac_ENSG00000267467; STARNET_eQTL_Mac_ENSG00000224916; STARNET_eQTL_Mac_ENSG00000234906; DeJager_Mic_ENSG00000130203; Kellis_Mic_ENSG00000130203; DeJager_Mic_ENSG00000130208; Kellis_Mic_ENSG00000130208 +# 4: AC_DeJager_eQTL_ENSG00000224916; AC_DeJager_eQTL_ENSG00000234906; BM_10_MSBB_eQTL_ENSG00000104853; Metabrain_Basalganglia_chr19_41840000_47960000_ENSG00000234906; Metabrain_Cerebellum_chr19_41840000_47960000_ENSG00000234906; Metabrain_Cortex_chr19_41840000_47960000_ENSG00000104853; Metabrain_Cortex_chr19_41840000_47960000_ENSG00000234906; Mic_DeJager_eQTL_ENSG00000130208; Mic_DeJager_eQTL_ENSG00000130203; Mic_mega_eQTL_ENSG00000130203; STARNET_eQTL_Mac_ENSG00000130208; STARNET_eQTL_Mac_ENSG00000130203; STARNET_eQTL_Mac_ENSG00000267467; STARNET_eQTL_Mac_ENSG00000224916; STARNET_eQTL_Mac_ENSG00000234906; DeJager_Mic_ENSG00000130203; Kellis_Mic_ENSG00000130203; DeJager_Mic_ENSG00000130208; Kellis_Mic_ENSG00000130208 +# 5: AC_DeJager_eQTL_ENSG00000189114; DLPFC_DeJager_eQTL_ENSG00000189114; Metabrain_Cortex_chr19_41840000_47960000_ENSG00000189114; PCC_DeJager_eQTL_ENSG00000189114; ROSMAP_AC_ENSG00000189114; ROSMAP_DLPFC_ENSG00000189114; ROSMAP_PCC_ENSG00000189114 +# 6: Ast_10_Kellis_eQTL_ENSG00000160007 + Ast_10_Kellis_eQTL_ENSG00000160007 +#set26 is APOE2? + +apoecs_clin2[cs_id=='set_26'&pos(variant_id2)==44908822]#no + +donors_apoecs[variant_id2=='chr19:44918487_G_T']$apoe2_r[1]#-0.21 +donors_apoecs[variant_id2=='chr19:44918487_G_T']$apoe4_r[1]#-0.23 + +#CCL: We can validate 3 of the 19 independant variants sets independently explaining one neuropathology after adjusted for APOE4 and 2 dose (along with Sex, education and age covariates), +#and 4/19 sets once we do not adjust for APOE4. These results is for FDR<0.05, using BH correction with n=19 tests + +# breakdown of the neuropathology associated sets: +# set27 is the APOE Mic eQTL that have previously been found associated with cerebral amyloid angiopathy in Fujita et al, and we are able to recapitulate this association. The 3 others sets are set_26, 31 and 57, +# respectively eQTL of APOC1 (in several brain regions and STARNET Mac), BLOC1S3( in several brain regions ), both explaining age at AD onset, and ARHGAP35 (in Ast.10 specifically), explaining Tau tangle level.