#########################################################################################
# Readme file for Age-AD, AD Cox Regression, and AD Linear Regression analyses:
# June 2021
#########################################################################################

#########################################################################################
# Overview
Using a cohort of 11,127 European-ancestry subjects with whole exome data (54% Alzheimer's Disease (AD) cases), a novel method of accounting for the effect of age on AD was tested in conjunction with more common methodologies. Details on the specific makeup of the discovery cohort used in the creation of the available summary stats are provided in Le Guen et al. 2021, "A novel age-informed approach for genetic association analysis in Alzheimer’s disease." Ancestry determination was completed using SNPWeightsv2.1 (Chen et al., 2013) with reference populations from the 1000 Genomes Consortium and all variants were filtered based on genotyping rate (<?95%), MAF?<?1% and Hardy-Weinberg equilibrium (HWE) in controls (p?<?10?6). Additionally, carriers of known pathogenic mutations on APP, PSEN1, PSEN2, and MAPT, as well as discordant pathology cases (any clinically diagnosed AD individual with Braak stage below III or neuritic plaques level below moderate) were excluded from our analysis. 

The goal of these analyses was to test a novel scoring method of integrating case-control status and age to better account for age-related AD changes in AD genetic studies. This novel AD-age score accounts for age by separating cases and controls by age in such a way that younger cases have higher scores than older cases, and older controls have lower scores than younger controls. This scoring methodology assumes that younger cases and older controls influence genetic associations more than older cases and younger controls, whose diagnostic status is more age-associated. Our analyses demonstrated that the AD-age score is a simple method to increase statistical power in AD genetic studies when compared to the more standard adjustment methods. 

Ten summary statistic files are provided. All files consists of SNP information and its association with AD diagnosis covaried for the first 10 prinicipal components and sex. SNP information is displayed in hg19. These ten summary statistic files represent five models for AD diagnosis with each model represent twice to include a second file that additionally covaries for APOE status. These models are:
1.	Logistic regression models estimated using PLINKv2.0 (Chang et al., 2015) with the –glm flag adjusted for age. 
2.	Logistic regression models estimated using PLINKv2.0 with the –glm flag unadjusted for age. 
3.	Cox regression models created using gwasurvir package (Rizvi et al., 2019) for R with regression on age at onset. 
4.	Linear regression on the AD-age score with a linear definition of age using PLINKv2.0 with the –glm flag.
5.	Linear regression on the AD-age score with a piecewise continuous definition of age using PLINKv2.0 with the –glm flag. 

More information regarding the analyses used to compute this data may be found in Le Guen et al. (2021), "A novel age-informed approach for genetic association analysis in Alzheimer’s disease." Current doi: 10.1186/s13195-021-00808-5.

#########################################################################################
# File Variables:
Files - 
•	AD_logistic_regression_Sex10PCs_p-value-only.tsv
•	AD_logistic_regression_Sex10PCs_Age_p-value-only.tsv
•	AD_logistic_regression_Sex10PCsAPOE2APOE4_p-value-only.tsv
•	AD_logistic_regression_Sex10PCsAPOE2APOE4_Age_p-value-only.tsv
Variables -
•	CHR: Chromosome
•	POS_hg19: Chromosome and base-pair coordinate
•	ID: The chromosome-position ID of the variant (Includes allele codes)
•	Effect_allele: The effect allele code of the variant
•	P-val: The p-value of the variant on the model for the file

Files - 
•	AD-age_score_linear_Sex10PCs_p-value-only.tsv
•	AD-age_score_linear_Sex10PCsAPOE2APOE4_p-value-only.tsv
•	AD-age_score_piecewise_Sex10PCsAPOE2APOE4_p-value-only.tsv
•	AD-age_score_piecewise_Sex10PCs_p-value-only.tsv
Variables -
•	CHR: Chromosome
•	POS_hg19: Chromosome and base-pair coordinate
•	ID: The chromosome-position ID of the variant (Includes allele codes)
•	Effect_allele: The effect allele code of the variant
•	P-val: The p-value of the variant on the model for the file
Files - 
•	Cox_regression_Sex10PCs_p-value-only.tsv
•	Cox_regression_Sex10PCsAPOE2APOE4_p-value-only.tsv
Variables -
•	CHR: Chromosome
•	POS_hg19: Chromosome and base-pair coordinate
•	ID: The chromosome-position ID of the variant (Includes allele codes)
•	Effect_allele: The effect allele code of the variant
•	P-val: The p-value of the variant on the model for the file

#########################################################################################
# References
Chen CY, Pollack S, Hunter DJ, Hirschhorn JN, Kraft P, Price AL. Improved ancestry inference using weights from external reference panels. Bioinformatics. 2013;29(11):1399–406. https://doi.org/10.1093/bioinformatics/btt144.
Chang CC, Chow CC, Tellier LC, Vattikuti S, Purcell SM, Lee JJ. Second-generation PLINK: rising to the challenge of larger and richer datasets. GigaSci. 2015;4(1):7. https://doi.org/10.1186/s13742-015-0047-8.
Rizvi AA, Karaesmen E, Morgan M, Preus L, Wang J, Sovic M, et al. gwasurvivr: an R package for genome-wide survival analysis. Bioinformatics. 2019;35:1968–70.