Studies	Statistical Analysis Descriptions	Figures/Tables
BIN1 isoform expression in DLPFC	There are FPKM values from RNAseq processing pipeline, unadjusted for covariates. Briefly, RNA was captured with next generation RNA-seq on the illumina HiSeq platform. Samples with RNA integrity score < 5 or quantity threshhold < 5ug we excluded, Fragments per kilobase were corrected for any batch effect with Combat. These FPKM values were used for analysis here in models further adjusting for demographic and technical covariates. Further details on this pipeline can be found elsewhere: pubmed/24508835; pubmed/26414614.	Fig.1c
Association of BIN1 isoforms with AD traits in DLPFC at RNA level	Linear regressions of neuropathology and cognitive characteristics (cognitive decline, MMSE, neuritic plaque, amyloid burden, tau burden and neurofibrillary tangles) versus 11 expressed isoforms of BIN1 in DLPFC derived using RNA-seq data. Models were adjusted for age at death, sex, cohort (ROS/MAP), and technical RNA processing covariates.	Suppl. Fig. 4
Measurement of BIN1 peptides abundance in DLPFC and its association with AD traits	Linear regressions of BIN1 peptides versus neuropathologies and cognitive characteristics (clinical AD, cognitive decline, residual cognition (adjusting for pathology), pathological AD diagnosis post mortem, amyloid burder, and tau burden), adjusting for age at death, sex, and cohort (ROS or MAP).	Fig. 4, Suppl. Table. 5
Association between BIN1 peptides and non-AD neuropathologic traits	Linear regressions of BIN1 peptides versus other neuropathologies (cerebral amyloid angiopathy, macro infarct, micro infart, lewy bodies, hippocampal sclerosis, and TDP), adjusting for age at death, sex, and cohort (ROS or MAP).	Suppl. Table. 6
Association of 7 BIN1 peptides with AD pathologies after adjusting for cell type proportion	Linear regressions of BIN1 peptides versus neuropathologies and cognitive characteristics (clinical AD, cognitive decline, residual cognition (adjusting for pathology), pathological AD diagnosis post mortem, amyloid burder, and tau burden), adjusting for age at death, sex, and cohort (ROS or MAP), as well as myeloid proportion, neuron proportion, astrocyte proportion, oligodendrocyte proportion, and endothelial cell proportion as estimated using DSA with RNA-seq.	Suppl. Table. 7
Association between BIN1 peptides encoded by exon 7 and tangles/amyloid after adjustment with tangle/amyloid burdens	Linear regressions of peptides versus tau burden, adjusting for age of death, sex, and amyloid burden, as well as linear regressions of peptides versus amyloid burden, adjusting for age, sex, and tau burden.	Fig. 5a, Suppl. Table. 8
Correlation between BIN1 exon 7/tangle burden/cognitive decline	Linear regressions of BIN1 peptides versus cognitive decline (which is adjusted for age at death, sex, and education in a random effects model) first without adjusting for tau, and then adjusting for tau. R2s reported are partial R2s.	Suppl. Table. 9, Fig. 5 b,c
Correlation between BIN1 exon 7/APOE4/cognitive decline	Linear regressions of BIN1 peptides, and APOEe4 allele count, versus cognitive decline (which is adjusted for age at death, sex, and education in a random effects model), modeled independently, as well as combined in a joint model. R2s reported are partial R2s.	Fig. 5 d, Table. 1
Study of the effect of AD susceptibility variants on BIN1 peptides in DLPFC	Linear regressions of BIN1 peptides versus IGAP SNPs (Lambert et. al., ) and Marioni family history of Alzheimer’s SNPs (Marioni et. al., ), adjusted for principal components.	Suppl. Table. 10
Evaluation of the effect of three BIN1 SNPs associated with AD on BIN1 isoform mRNA expression in DLPFC and in purified microglia	Generalized linear regression was conducted to analyze the association between the genotype and expression level with the adjustment of the age at death, sex, post mortem interval, and major ethnicity principal components	Suppl. Fig. 3 b, c
Evaluation of the effect of BIN1 SNPs associated with AD identified by GWAS on BIN1 isoform expression at the mRNA level in DLPFC	The analysis included the filtered 898 single nucleotide polymorphisms (SNPs) (imputation quality ≥ 0.9, minor allele count ≥ 10, and minor allele frequency ≥ 0.05) within the genomic region of ±100 Kb region close to BIN1 (chr2: 127,705,602-127,964,931, Build hg19). The number of SNPs with Bonferroni corrected significance (p<0.05/898=5.57E-5) and nominal significance (p<0.05) were presented. The analysis was adjusted for age at death, sex, and major ethnic principal components.	Suppl. Table. 4 a, b
Association of the BIN1 peptide LQAHLVAQTNLLR encoded by exon 7 with tangles	Linear regressions of tau burden versus BIN1 peptides, adjusting for age at death, sex, as well as miRNA, and DLPFC modules, based on RNA-seq, which were previously shown to be associated with tau burden in this cohort (ROS/MAP).	Suppl. Table. 11
Study of association between Tau-related epigenomic alterations and BIN1 peptides.	Linear regressions of BIN1 peptides versus ETES (Epigenomic Tau Effect Score), adjusting for age at death, and sex.	Suppl. Table. 12