Merge pull request #83 from uclahs-cds/nzeltser-add-auc-plotting

add auc plotting

Author: alkaZeltser

DESCRIPTION

@@ -27,6 +27,7 @@ Imports:
 Suggests:
     knitr,
     rmarkdown,
+    scales,
     testthat (>= 3.0.0)
 Config/testthat/edition: 3
 License: GPL-2

NAMESPACE

@@ -18,16 +18,21 @@ importFrom(
     'colorRampPalette',
     'col2rgb'
     );
+
 importFrom(
     'stats',
     'aggregate',
     'as.formula',
     'binomial',
+    'coef',
+    'complete.cases',
     'glm',
     'lm',
     'na.omit',
+    'pnorm',
     'predict',
     'quantile',
+    'vcov',
     'setNames'
     );
 importFrom(
@@ -50,6 +55,7 @@ export(
     import.pgs.weight.file,
     import.vcf,
     parse.pgs.input.header,
+    analyze.pgs.binary.predictiveness,
     create.pgs.density.plot,
     create.pgs.boxplot,
     create.pgs.rank.plot,

NEWS.md

@@ -12,6 +12,7 @@
 * Added new plotting function `create.pgs.boxplot`
 * Added option for user to provide custom PGS source column(s) for plotting functions
 * Added option to `assess.pgs.vcf.allele.match` to condition the handling of ambiguous strand flips on the total number of unambiguous strand flips.
+* Added new function `analyze.pgs.binary.predictiveness` which given a PGS and phenotypes runs a logistic regression and returns statistics (OR, p-value, AUC) describing how well the PGS predicts the phenotype. It also automatically plots a receiver-operator-curve.
 
 # ApplyPolygenicScore 3.0.2