MM-136 another one

R/binary_to_liability.R

@@ -1,8 +1,9 @@
-#' Convert binary traits to liabilities
+#' Convert binary traits to liabilities for low prevalence
 #'
-#' This function implements the conversion of binary traits to liabilties as
+#' @description
+#' This function implements an approximate conversion of binary traits to liabilties as
 #' proposed in the LT-MAPIT model (Crawford and Zhou 2018,
-#' https://doi.org/10.1101/374983).
+#' https://doi.org/10.1101/374983). Note that this is only good for low prevalence. 
 #' To run LT-MAPIT (MAPIT on case-control traits), convert the binary traits to
 #' liabilities using this function and pass the liabilities to mvmapit as trait.
 #'

vignettes/tutorial-lt-mapit.Rmd

@@ -44,7 +44,6 @@ sample_size <- 500
 ## Simulate random genotypes
 
 Simulate the genotypes such that all variants have minor allele frequency (MAF) > 0.05.
-**NOTE:** As in the paper, we center and scale each genotypic vector such that every SNP has mean 0 and standard deviation 1.
 
 ```{r random_genotypes, eval = F}
 maf <- 0.05 + 0.45 * runif(n_snps)
@@ -90,6 +89,7 @@ X <- simulated_data$genotype[c(cases, controls), ]
 
 To run MAPIT with case-control traits, we need to convert the traits back to liabilities.
 The function `binary_to_liability` provides this conversion.
+**NOTE:** The `binary_to_liability` function is an approximation that is only suited for low prevalence in the disease trait.
 
 ```{r mapit, eval = F}
 y_liabilities <- binary_to_liability(y, disease_prevalence)