multi_resids: Get the residuals from multiple linear models.
In mirvie/mirmodels: Models Built by the Informatics Team at Mirvie

Description Usage Arguments Value Examples

There is the option to add the mean (or median for robust models) to the residuals.

1	multi_resids(lms, lms_data = NULL, reset_mean_med = FALSE, new_data = NULL)

`lms`	A list of fitted linear models (the results of calls to `stats::lm()` or `MASS::rlm()`). Most likely, the output of a call to `multi_lm()`.
`lms_data`	The data frame that was passed as the `df` argument to `multi_lm()`, if `multi_lm()` was used to create `lms`.
`reset_mean_med`	A flag. If `TRUE`, for each variable for which the residuals are calculated the mean (or in the case of robust linear models, the median) of the original values of these variables is added to the residuals, such that the output is again centred on this value.
`new_data`	Rather than calculating the residuals on the data where the model was fit, you can pass a new dataset and calculate the residuals there.

A data frame of the residuals. The variables in this data frame will have their original names. If lms_data is given, other columns (ones for which residuals were not calculated) of that data frame will be returned as is.

if (rlang::is_installed("mirmisc")) {
  gars_data <- get_combined_cohort_data(c("ga", "rs"), log2 = TRUE) %>%
    dplyr::mutate(
      cohort = dplyr::if_else(startsWith(mirvie_id, "RS"), "RS", "GA"),
      cohort = factor(cohort),
      meta_major_race = forcats::fct_drop(meta_major_race)
    ) %>%
    dplyr::filter(!is.na(meta_major_race), !is.na(meta_collectionga))
  xs <- c("cohort", "meta_major_race", "meta_collectionga")
  ys <- gars_data %>%
    dplyr::select(dplyr::any_of(mirmisc::get_gene_names())) %>%
    purrr::map_dbl(sum) %>%
    sort() %>%
    tail(100) %>%
    names() # ys are the highest expressed 100 genes
  lms <- multi_lm(gars_data, xs, ys)
  resids <- multi_resids(lms, gars_data, reset_mean_med = TRUE)
}