owas_qgcomp {epiomics}

Perform omics wide association study using qgcomp

Description

Omics wide association study using quantile-based g-Computation (as described by Keil et al., (2019) doi:10.1289/EHP5838) to examine associations of exposure mixtures with each individual 'omics feature as an outcome 'omics data as either the dependent variable. Allows for either continuous or dichotomous outcomes, and provides the option to adjust for covariates.

Usage

owas_qgcomp(
  df,
  expnms,
  omics,
  covars = NULL,
  q = 4,
  confidence_level = 0.95,
  family = "gaussian",
  rr = TRUE,
  run.qgcomp.boot = TRUE
)

Arguments

df

Dataset

expnms

Name of the exposures. Can be either continuous or dichotomous. For dichotomous variables, must set q to "NULL", and values must be either 0/1.

omics

Names of all omics features in the dataset

covars

Names of covariates (can be NULL)

q

NULL or number of quantiles used to create quantile indicator variables representing the exposure variables. Defaults to 4If NULL, then qgcomp proceeds with un-transformed version of exposures in the input datasets (useful if data are already transformed, or for performing standard g-computation).

confidence_level

Confidence level for marginal significance (defaults to 0.95, or an alpha of 0.05)

family

Currently only "gaussian" (default) for linear models (via lm) or "binomial" for logistic. Default is "gaussian".

rr

see qgcomp()

run.qgcomp.boot

Should the model be fit with qgcomp.boot? See package qgcomp.boot for details. Default is TRUE. Setting to FALSE decreases computational time.

Value

A data frame with the following columns: feature: name of the omics feature psi: the model estimate for the feature. For linear models, this is the beta; for logistic models, this is the log odds. lcl_psi: the lower confidence interval. ucl_psi: the upper confidence interval. p_value: p-value for the estimate test_statistic: t-statistic for psi coefficient adjusted_pval: FDR adjusted p-value threshold: Marginal significance, based on unadjusted p-values covariates: the names of covariates in the model, if any coef_exposure: the individual coefficient of each exposure

Examples

# Load Example Data
data("example_data")

# Get names of omics
colnames_omic_fts <- colnames(example_data)[grep("feature_",
                                              colnames(example_data))][1:5]

# Names of exposures in mixture
 exposure_names = c("exposure1", "exposure2", "exposure3")

# Run function without covariates
out <- owas_qgcomp(df = example_data,
                   expnms = exposure_names,
                   omics = colnames_omic_fts,
                   q = 4, 
                   confidence_level = 0.95) 


# Run analysis with covariates
out <- owas_qgcomp(df = example_data,
                   expnms = c("exposure1", "exposure2", "exposure3"),
                   covars = c("weight", "age", "sex"),
                   omics = colnames_omic_fts,
                   q = 4, 
                   confidence_level = 0.95)