HIMA (version 2.2.2)

hima2: High-dimensional Mediation Analysis

Description

hima2 is a wrapper function designed to perform various HIMA methods for estimating and testing high-dimensional mediation effects. hima2 can automatically select the appropriate HIMA method based on the outcome and mediator data type specified by the user.

Usage

hima2(
  formula,
  data.pheno,
  data.M,
  outcome.family = c("gaussian", "binomial", "survival", "quantile"),
  mediator.family = c("gaussian", "negbin", "compositional"),
  penalty = c("DBlasso", "MCP", "SCAD", "lasso"),
  efficient = FALSE,
  scale = TRUE,
  Sigcut = 0.05,
  verbose = FALSE,
  ...
)

Value

A data.frame containing mediation testing results of selected mediators.

ID:

Mediator ID/name.

alpha:

Coefficient estimates of exposure (X) --> mediators (M) (adjusted for covariates).

beta:

Coefficient estimates of mediators (M) --> outcome (Y) (adjusted for covariates and exposure).

alpha*beta:

Mediation (indirect) effect.

Relative Importance:

Relative importance of the mediator. It is the proportion of the mediation effect for each mediator out of the sum of the mediation effect (absolute value) across all significant mediators selected.

p-value:

Joint raw p-value of significant mediators selected based on corresponding approach.

tau:

Quantile level of the outcome (applicable only to the quantile mediation model).

Arguments

formula

an object of class formula: a symbolic description of the overall effect model, i.e., outcome ~ exposure + covariates, to be fitted. Make sure the "exposure" is the variable of interest, which must be listed as the first variable in the right hand side of the formula.

data.pheno

a data frame containing exposure and covariates that are listed in the right hand side of the formula. The variable names must match those listed in formula. By default hima2 will scale data.pheno.

data.M

a data.frame or matrix of high-dimensional mediators. Rows represent samples, columns represent variables. By default hima2 will scale data.M.

outcome.family

either 'gaussian' (default, for normally distributed continuous outcome), 'binomial' (for binary outcome), 'survival' (for time-to-event outcome), or 'quantile' (for quantile mediation analysis)

mediator.family

either 'gaussian' (default, for continuous mediators), 'negbin' (i.e., negative binomial, for RNA-seq data as mediators), or 'compositional' (for microbiome data as mediators), depending on the data type of high-dimensional mediators (data.M).

penalty

the penalty to be applied to the model. Either 'DBlasso' (De-biased LASSO, default), 'MCP', 'SCAD', or 'lasso'. Please note, survival HIMA and microbiome HIMA can be only performed with 'DBlasso'; Quantile HIMA cannot be performed with 'DBlasso'. Not applicable for efficient HIMA (when efficient = TRUE), as it will always apply 'MCP'.

efficient

use efficient HIMA (eHIMA). Only applicable for linear HIMA with continuous outcome (i.e., outcome.family = "gaussian" and mediator.family = "gaussian"). Default = FALSE.

scale

logical. Should the function scale the data (exposure, mediators, and covariates)? Default = TRUE. Note: for simulation study, scale can be turned off to avoid estimate compression.

Sigcut

cutoff applied to select significant mediators. Default = 0.05.

verbose

logical. Should the function be verbose and show the progression? Default = FALSE.

...

reserved passing parameter.

References

1. Zhang H, Zheng Y, Zhang Z, Gao T, Joyce B, Yoon G, Zhang W, Schwartz J, Just A, Colicino E, Vokonas P, Zhao L, Lv J, Baccarelli A, Hou L, Liu L. Estimating and Testing High-dimensional Mediation Effects in Epigenetic Studies. Bioinformatics. 2016. DOI: 10.1093/bioinformatics/btw351. PMID: 27357171; PMCID: PMC5048064

2. Zhang H, Zheng Y, Hou L, Zheng C, Liu L. Mediation Analysis for Survival Data with High-Dimensional Mediators. Bioinformatics. 2021. DOI: 10.1093/bioinformatics/btab564. PMID: 34343267; PMCID: PMC8570823

3. Zhang H, Chen J, Feng Y, Wang C, Li H, Liu L. Mediation Effect Selection in High-dimensional and Compositional Microbiome data. Stat Med. 2021. DOI: 10.1002/sim.8808. PMID: 33205470; PMCID: PMC7855955

4. Zhang H, Chen J, Li Z, Liu L. Testing for Mediation Effect with Application to Human Microbiome Data. Stat Biosci. 2021. DOI: 10.1007/s12561-019-09253-3. PMID: 34093887; PMCID: PMC8177450

5. Perera C, Zhang H, Zheng Y, Hou L, Qu A, Zheng C, Xie K, Liu L. HIMA2: High-dimensional Mediation Analysis and Its Application in Epigenome-wide DNA Methylation Data. BMC Bioinformatics. 2022. DOI: 10.1186/s12859-022-04748-1. PMID: 35879655; PMCID: PMC9310002

6. Zhang H, Hong X, Zheng Y, Hou L, Zheng C, Wang X, Liu L. High-Dimensional Quantile Mediation Analysis with Application to a Birth Cohort Study of Mother–Newborn Pairs. Bioinformatics. 2024. DOI: 10.1093/bioinformatics/btae055. PMID: 38290773; PMCID: PMC10873903

7. Bai X, Zheng Y, Hou L, Zheng C, Liu L, Zhang H. An Efficient Testing Procedure for High-dimensional Mediators with FDR Control. Statistics in Biosciences. 2024. DOI: 10.1007/s12561-024-09447-4.

Examples

if (FALSE) {
# Note: In the following examples, M1, M2, and M3 are true mediators.
data(himaDat)

# Example 1 (continous outcome - linear hima): 
head(himaDat$Example1$PhenoData)

e1 <- hima2(Outcome ~ Treatment + Sex + Age, 
      data.pheno = himaDat$Example1$PhenoData, 
      data.M = himaDat$Example1$Mediator,
      outcome.family = "gaussian",
      mediator.family = "gaussian",
      penalty = "MCP", # Can be "DBlasso" for dblassoHIMA
      scale = FALSE) # Disabled only for simulation data
e1
attributes(e1)$variable.labels

# Efficient HIMA (only applicable to mediators and outcomes that are 
# both continuous and normally distributed.)
e1e <- hima2(Outcome ~ Treatment + Sex + Age,
             data.pheno = himaDat$Example1$PhenoData,
             data.M = himaDat$Example1$Mediator,
             outcome.family = "gaussian",
             mediator.family = "gaussian",
             efficient = TRUE,
             scale = FALSE) # Disabled only for simulation data
e1e
attributes(e1e)$variable.labels

# Example 2 (binary outcome - logistic hima): 
head(himaDat$Example2$PhenoData)

e2 <- hima2(Disease ~ Treatment + Sex + Age, 
      data.pheno = himaDat$Example2$PhenoData, 
      data.M = himaDat$Example2$Mediator,
      outcome.family = "binomial",
      mediator.family = "gaussian",
      penalty = "MCP",
      scale = FALSE) # Disabled only for simulation data
e2
attributes(e2)$variable.labels

# Example 3 (time-to-event outcome - survival hima): 
head(himaDat$Example3$PhenoData)

e3 <- hima2(Surv(Status, Time) ~ Treatment + Sex + Age, 
      data.pheno = himaDat$Example3$PhenoData, 
      data.M = himaDat$Example3$Mediator,
      outcome.family = "survival",
      mediator.family = "gaussian",
      penalty = "DBlasso",
      scale = FALSE) # Disabled only for simulation data
e3
attributes(e3)$variable.labels

# Example 4 (compositional data as mediator, e.g., microbiome): 
head(himaDat$Example4$PhenoData)

e4 <- hima2(Outcome ~ Treatment + Sex + Age, 
      data.pheno = himaDat$Example4$PhenoData, 
      data.M = himaDat$Example4$Mediator,
      outcome.family = "gaussian",
      mediator.family = "compositional",
      penalty = "DBlasso") # Scaling is always enabled for microHIMA
e4
attributes(e4)$variable.labels

#' # Example 5 (quantile mediation anlaysis - quantile hima): 
head(himaDat$Example5$PhenoData)

# Note that the function will prompt input for quantile level.
e5 <- hima2(Outcome ~ Treatment + Sex + Age, 
      data.pheno = himaDat$Example5$PhenoData, 
      data.M = himaDat$Example5$Mediator,
      outcome.family = "quantile",
      mediator.family = "gaussian",
      penalty = "MCP", # Quantile HIMA does not support DBlasso
      scale = FALSE, # Disabled only for simulation data
      tau = c(0.3, 0.5, 0.7)) # Specify multiple quantile level
e5
attributes(e5)$variable.labels
}