README file for p_ACT_meta.R 
P-values Adjusted for Correlated Tests, meta-analysis version

***Note: this software is available as a beta version.  It is based on the 
method of Conneely and Boehnke (2010 - currently in revision at Genetic 
Epidemiology).  Since this is a beta version based on a yet unpublished article,
feel free to contact kconnee@emory.edu for additional information if you are 
interested in using this software.

p_ACT_meta.R is a post-processing program for meta-analyses that adjusts 
for multiple testing of candidate gene studies in a way that accounts for 
correlation between SNPs in the same gene.  

To run this program, R (http://www.r-project.org/) must be installed, as 
well as the mvtnorm (version > 0.8-0) and corpcor packages.

Programs may be run interactively in UNIX and Windows environments by 
typing 'source(p_ACT.R)' from within an R session. In UNIX environments, 
batch mode is recommended; simply type 'R CMD BATCH p_ACT.R' at the prompt.

Output file: p_ACT_meta.out 

Input files: zstats.csv, weights.csv, genelist.csv (if multiple genes are 
  involved)), two_stage.csv (if a two-stage analysis was performed), and 
  geno1.csv-genoJ.csv, where J is the number of studies combined
  in the meta-analysis.
  
Input file formats: CSV files (can be easily created with Excel).

two_stage.csv
   This file should only be included if a two-stage analysis has been performed.  It should
   consist of a single numeric entry representing the p-value cutoff used to determine 
   whether a SNP is followed up in stage 2.
Example:
0.1

genelist.csv
   This file should only be included if the meta-analyses includes multiple genes.  If this
   file is not present, a single-gene analysis will be assumed.  File should include 1 row 
   for each SNP, and two columns: 1) Gene name 2) SNP name.
Example:
Gene_A,rs1
Gene_A,rs2
Gene_B,rs3
Gene_B,rs4
Gene_B,rs5


The zstats.csv and weights.csv files are required for all analyses.  They should include 
values of Z-statistics and appropriate weights as shown in Table 2 of Conneely and Boehnke 
(2010 - currently unpublished, a link will be added when available.)

zstats.csv
   File should contain one row for each SNP, and J+1 columns (J = # of studies).  
   Column 1: SNP names
   Columns 2 - J+1: Z-statistics from each study in the meta-analysis
   Missing values should be designated NA
   If a two-stage analysis is performed, it will be assumed that Column 2 contains
     the initial stage and Columns 3-J+1 are analyses in the follow-up stage.  All
     SNPs not included in the follow-up stage should have a z-statistic of NA. 
Example:
rs1,-2.99986,-2.13205
rs2,-2.99839,-0.22161
rs3,-0.4893,-0.66602
rs4,-0.35492,-0.65514
rs5,-0.48716,NA

weights.csv
   File should contain one row for each SNP, and J+1 columns (J = # of studies).  
   Column 1: SNP names
   Columns 2 - J+1: Weights for each study in the meta-analysis
   Missing values (and weights for SNPs not followed up in subsequent studies in a two-stage
      analysis) should be designated NA or 0. 
rs1,0.6588,0.7524
rs2,0.6606,0.7508
rs3,0.6588,0.7524
rs4,0.6606,0.7508
rs5,1,0

geno1.csv - genoJ.csv
   For two stage analyses, only geno1.csv should be present.  For all other analyses, J files are
   needed, where J is the number of studies combined in the meta-analysis.  File j should contain
   1 row for each SNP, and N(j)+1 columns, where N(j) is the number of individuals in study j.
   Column 1: SNP names
   Columns 2-N(j): genotype scores for all individuals in each study. (These will often be allele
     counts - ie 0,1,2.)  For examples, see included files in Sample_files directory.

The directories Sample_files and Sample_files_2s contain example files for a conventional meta-analysis
and a two-stage analysis, respectively.