Data for "Brain cell-type shifts in Alzheimer's disease, autism and schizophrenia interrogated using methylomics and genetics"

Data for "Genetic and methylomic interrogation of brain cell-type shifts in autism, schizophrenia, and Alzheimer’s disease" (Yap et al. 2023).

Source data from ROSMAP, LIBD and UCLA_ASD post-mortem brain datasets.

This data repository contains 3 files:

220819_Supplementary_Tables.xlsx

Supplementary Tables for the manuscript:

1. Supplementary Table 1: Comparison of CTP deconvolution methods in the ROSMAP dataset. mcc* denotes methylCC deconvolution; sSV* denotes smartSVA; h* indicates Houseman using array reference data; hseq* indicates Houseman with sequencing reference data; celfie* indicates CelFIE (and includes an output titled "unknown1"); VAEe* indicates variational autoencoder embeddings.
2. Supplementary Table 2: Comparison of CTP deconvolution methods in the LIBD dataset. mcc* denotes methylCC deconvolution; sSV* denotes smartSVA; h* indicates Houseman using array reference data; hseq* indicates Houseman with sequencing reference data; celfie* indicates CelFIE (and includes an output titled "unknown1"); VAEe* indicates variational autoencoder embeddings.
3. Supplementary Table 3: Comparison of CTP deconvolution methods in the UCLA_ASD dataset. mcc* denotes methylCC deconvolution; sSV* denotes smartSVA; h* indicates Houseman using array reference data; hseq* indicates Houseman with sequencing reference data; celfie* indicates CelFIE (and includes an output titled "unknown1"); VAEe* indicates variational autoencoder embeddings.
4. Supplementary Table 4: Deconvolved brain CTPs (raw), with covariates.
5. Supplementary Table 5: Deconvolved brain CTPs (clr-transform, offset 1e-3), with covariates and raw PGS for all ancestries. Comp* indicates compositionally-aware principal components of the CTP data; *pgs_raw indicates PGS calculated for using genotyping across all ancestries. This table has a total of n=1,098 across all ancestries, including n=885 EUR. After applying a rel<0.05 threshold on the n=885 EUR, there were n=878 EUR which were used in the PGS analysis so that population stratification PCs did not simply capture family structure.
6. Supplementary Table 6: Deconvolved brain CTPs (clr-transform, offset 1e-3), with covariates and standardised PGS for n=878 Europeans. *pgs_raw indicates unstandardised PGS, PC* indicates genotyping PCs within the European dataset, *_PGS indicates standardised PGS within the European subset.
7. Supplementary Table 7: Deconvolved brain CTPs (clr-transform, offset 1e-3), adjusted for oligodendrocyte proportions, with covariates.
8. Supplementary Table 8: Deconvolved brain CTPs (raw), adjusted for oligodendrocyte proportions, with covariates.

20220108_maf05_gwas_ctp.tar.gz

GWAS summary statistics for the 7 brainCTPs (clr-transformed): Exc, Inh, Astro, Endo, Micro, Oligo, OPC

METAL output format:
MarkerName: SNP
Allele1
Allele2
Freq1: Allele1 frequency
FreqSE: frequency standard error
MinFreq: minimum Allele1 frequency in meta-analysis
MaxFreq: maximum Allele1 frequency in meta-analysis
Effect: effect size
StdErr: standard error of effect size
P-value: calculated in inverse variance weighted meta-analysis
Direction: directions of effects across the 3 datasets
HetISq: heterozygosity I-squared
HetChiSq: heterozygosity chi-squared
HetDf: heterozygosity degress of freedom
HetPVal: heterozygosity test p-value

20220108_maf05_gwas_ctp_pc.tar.gz

GWAS summary statistics for the 5 CTP_PCs

METAL output format (see above)

Source data:

ROSMAP: Raw methylation .idat files were obtained from Synapse accession syn7357283. Whole genome sequencing .vcf files (variants jointly called with MSBB and Mayo studies) were obtained from Synapse accession syn11707420.

LIBD: Raw methylation .idat files were obtained from GEO accession GSE74193. SNP genotypes were downloaded from dbGaP accession phs000417.v2.p1.

UCLA-ASD: The processed methylation beta matrix was downloaded from Synapse accession syn8263588. SNP genotypes were downloaded from Synapse accession syn10537134.

GWAS summary statistics are available at: 10.5281/zenodo.7604231

Code is available on GitHub: gandallab/brain_CTP_deconv