Gather lists of neighbors of true ones from the graph and detected ones from cv.glmnet()

Usage

final_step(df_truth, df_glm, robustness_step = NULL)

Arguments

df_truth: Dataframe. The one resulting from truth_by_prevalence()
df_glm: Dataframe. The one resulting from cvglm_to_coeffs_by_object()
robustness_step: Boolean. When TRUE, filtering_top will be different from 100%, when FALSE the reults from the naïve method are looked at

Value

Dataframe. Returns for each level of prevalence and module ID, the list of true and/or detected neighbors and the corresponding list of coefficients

Examples

# Dataframe with true neighbors
df_true <- list(
  tibble::tibble(
    node1 = c("msp_1", "msp_1", "msp_2", "msp_3"), node2 = c("msp_55", "msp_20", "msp_3", "msp_18"),
    prev1 = c(0.28, 0.28, 0.96, 0.75), prev2 = c(0.76, 0.25, 0.75, 0.60)
  ),
  tibble::tibble(node1 = c("msp_2", "msp_3"), node2 = c("msp_3", "msp_18"), prev1 = c(0.96, 0.75), prev2 = c(0.75, 0.60))
) %>% rlang::set_names(c("0.20", "0.30"))

# Dataframe with detected neighbors
df_detected <- list(
  tibble::tibble(
    prev_level = c("0.20", "0.30", "0.30", "0.30"), node1 = c("msp_2", "msp_2", "msp_3", "msp_3"),
    node2 = c("msp_3", "msp_3", "msp_18", "msp_8"), coef = c(0.406, -0.025, 0.160, 0.005),
    filtering_top = c(100, 100, 100, 100)
  ),
  tibble::tibble()
) %>% rlang::set_names(c("0.20", "0.30"))
# Use final_step() to gather both
neighbors <- final_step(df_true, df_detected, robustness_step = FALSE)