search

PankratzLab / GenScorePipeline

Genetic Score Pipeline, Merge Extract Pipeline
1 stars 0 forks source link

readme

Gen Score Pipeline

Note: run org.genvisis.cnv.gwas.utils.GeneScorePipeline -h to see what is required to run the pipeline.

Prior to analysis

  1. Create files:

    • Genetic dataset

    • data.txt file that points to genetic dataset (tab-delimited file – D label location of data/map files extension of data files extension of map files full path to IDs file [any file; the first two columns are read as IDs])

    • Effect files saved as .meta files

      • Meta files must contain ‘MarkerName’, ‘Effect’, ‘Freq’, ‘Pvalue’, ‘Chr’ and ‘Position’ (MarkerName should be reflective of name in the genetic dataset) (column names are not fixed - see appendix for notes on usable column names).
      • If ‘Chr’ and ‘Position’ are not present, there is a pre-process step available to look those up using DBSnp databases. To obtain chr and position for a list of rsIDs go to http://genreport.umn.edu/#topOfPage
      • For a single SNP the effect (beta value) should be 1.

    • .pheno files

      • The first 3 columns should be ‘fid’, ‘iid’, and ‘case_control’
      • Then add columns for any covariates (e.g., age, sex, etc.)

  2. Create a location directory for GSP (e.g., GSP)

    • Make a Data directory inside the GSP directory
      • Place .pheno files and a data.txt file here
    • Place .meta files inside the GSP location directory

Run GenScorePipeline

Run GSP from the directory above the GSP directory.

Review your results

Result Column Explanation
STUDY study name
DATAFILE Data file used for results line
INDEX-THRESHOLD
FACTOR .pheno filename used for results line
BASE-R-SQUARED r2 for model ran using genetics only no covariates
R-SQR r2 for model ran using genetics with covariates
R-DIFF difference between the two r2; larger is better
P-VALUE from software
EXCEL-SIG p-value calculated in excel
BETA effect estimate for the risk score
SE standard error of the effect estimate
NUM total number used in analysis
CASES total number of cases used in the analysis (if continuous pheno will be NA)
CONTROLS total number of controls used in the analysis (if continuous pheno will be NA)
PAIRED-T-P-VALUE used with TRIOS; average of parent compared to child (N/A if not a trio)
WILCOXON-SIGNED-RANK-P-VALUE used with TRIOS; average of parent compared to child (N/A if not a trio)
PAIRED-STAT-NUM-TRIOS used with TRIOS; average of parent compared to child (0 if not a trio)
#sigInMeta <5x10-08 (number of variants in the metafile that were below the 5x10^-8 threshold [or whatever threshold you specified in the command])
#indexVariantsInMeta Number we have in our dataset
#indexVariantsInDataset Number we have in our dataset
B-F-SCORE How much is captured if T < U : missingness; ideal =1
INVCHI-SCORE How much is captured if T < U : missingness; ideal =1

Appendix 1: Column Names

Column names can be any from a list of allowed aliases, e.g., ‘SNP’ instead of ‘MarkerName’. Column aliases are defined in the Genvisis source code in the org.pankratzlab.common.Aliases class, using: MARKER_NAMES CHRS POSITIONS EFFECTS ALLELE_FREQS PVALUES STD_ERRS