In insightsengineering/hermes: Preprocessing, analyzing, and reporting of RNA-seq data

Objective: replace tern::df_explicit_na with a simpler hermes function

For the example, we use the colData from the summarized_experiment object in the hermes package. We add one missing value to the categorical variable AGE18, add one missing value to the continuous variable BAGE, and one missing value to the logical variable LowDepthFlag.

library(hermes)
dat <- colData(summarized_experiment)
dat[1, 4] <- NA
dat[1, 59] <- NA
dat[1, 86] <- NA

Raw Code

We use forcats::fct_explicit_na() function to add a factor level (Missing). This skips using the df_explicit_na functions in tern.

We use lapply(, factor) to convert character variables into factors without using loop, which is used in tern::df_explicit_na.

hermes_explicit_na <- function(data, na_level = "<Missing>") {
    assert_that(is(data, "DataFrame"))

    var_is_logical <- sapply(data, is.logical)
    data[,var_is_logical] <- lapply(data[,var_is_logical], as.character)

    var_is_character <- sapply(data, is.character)
    data[,var_is_character] <- lapply(data[,var_is_character], factor)

    var_has_missing <- sapply(data, anyNA)
    var_to_add_NA_level <- var_is_character & var_has_missing
    data[,var_to_add_NA_level] <- lapply(data[,var_to_add_NA_level], 
                                         function(s) forcats::fct_explicit_na(s, 
                                                            na_level = na_level)       
                                         )
    data

}

Test

A new level (Missing) is added to AGE18.

dat_hermes_convert <- hermes_explicit_na(dat)

dat_hermes_convert$AGE18

Continuous variable BAGE does not have any change.

dat_hermes_convert$BAGE

Logical variable is converted to factor. A new level (Missing) is added.

dat_hermes_convert$LowDepthFlag

Original variable with all records missing is converted to a factor with one level (Missing).

dat_hermes_convert$RACE3

insightsengineering/hermes documentation built on March 11, 2024, 11:04 p.m.