| Title: | A Zero-Inflated Beta Random Effect Model |
|---|---|
| Description: | A two-part zero-inflated Beta regression model with random effects (ZIBR) for testing the association between microbial abundance and clinical covariates for longitudinal microbiome data. Eric Z. Chen and Hongzhe Li (2016) <doi:10.1093/bioinformatics/btw308>. |
| Authors: | Eric Zhang Chen [aut, cph], Charlie Bushman [cre] |
| Maintainer: | Charlie Bushman <[email protected]> |
| License: | MIT + file LICENSE |
| Version: | 1.0.2.9000 |
| Built: | 2025-02-11 05:14:00 UTC |
| Source: | https://github.com/pennchopmicrobiomeprogram/zibr |
Fit beta random effect
fit_beta_random_effect( Z = Z, Y = Y, subject.ind = subject.ind, time.ind = time.ind, quad.n = 30, verbose = FALSE )fit_beta_random_effect( Z = Z, Y = Y, subject.ind = subject.ind, time.ind = time.ind, quad.n = 30, verbose = FALSE )
Z |
FILL |
Y |
FILL |
subject.ind |
the subject index |
time.ind |
the time index |
quad.n |
number of points in gaussian quadrature |
verbose |
a boolean to enable more output |
a named list
est.table
s2.est
v.est
Fit logisitic random effect
fit_logistic_random_effect( X = X, Y = Y, subject.ind = subject.ind, time.ind = time.ind, quad.n = 30, verbose = FALSE )fit_logistic_random_effect( X = X, Y = Y, subject.ind = subject.ind, time.ind = time.ind, quad.n = 30, verbose = FALSE )
X |
FILL |
Y |
FILL |
subject.ind |
the subject index |
time.ind |
the time index |
quad.n |
number of points in gaussian quadrature |
verbose |
a boolean to enable more output |
a named list
est.table
s1.est
Fit zero inflated beta random effect
fit_zero_inflated_beta_random_effect( X = X, Z = Z, Y = Y, subject_ind = subject_ind, time_ind = time_ind, component_wise_test = TRUE, joint_test = TRUE, quad_n = 30, verbose = FALSE )fit_zero_inflated_beta_random_effect( X = X, Z = Z, Y = Y, subject_ind = subject_ind, time_ind = time_ind, component_wise_test = TRUE, joint_test = TRUE, quad_n = 30, verbose = FALSE )
X |
FILL |
Z |
FILL |
Y |
FILL |
subject_ind |
the subject index |
time_ind |
the time index |
component_wise_test |
boolean to run component-wise test |
joint_test |
boolean to run joint test |
quad_n |
number of points in gaussian quadrature |
verbose |
a boolean to enable more output |
a named list
logistic_est_table
logistic_s1_est
beta_est_table
beta_s2_est
beta_v_est
loglikelihood
joint_p
A dataset containing the bacterial abundance and clinical information from a longitudinal human microbiome study
ibdibd
A data frame with 236 rows and 5 variables:
Sample IDs
Subject IDs
Time points
Treatment, 0 for antiTNF, 1 for EEN
Abundance for Eubacterium
...
Lewis and Chen et al. (2016) Cell Host & Microbe 18 (4), 489-500
Simulate beta data
simulate_beta_random_effect_data( subject_n = 50, time_n = 5, v = 2, beta = as.matrix(c(-0.5, -0.5, 0.5)), Z = NA, s2 = 1, sim_seed = 100 )simulate_beta_random_effect_data( subject_n = 50, time_n = 5, v = 2, beta = as.matrix(c(-0.5, -0.5, 0.5)), Z = NA, s2 = 1, sim_seed = 100 )
subject_n |
the number of subjects |
time_n |
the number of time points |
v |
FILL |
beta |
FILL |
Z |
FILL |
s2 |
FILL |
sim_seed |
the random seed with which to simulate the data |
a named list
Y
Z
c
u
v
beta
s2
subject_ind
time_ind
Simulate logistic data
simulate_logistic_data( subject_n = 50, time_n = 5, alpha = as.matrix(c(0, 0.5, -1)), s1 = 0.5, sim_seed = 100 )simulate_logistic_data( subject_n = 50, time_n = 5, alpha = as.matrix(c(0, 0.5, -1)), s1 = 0.5, sim_seed = 100 )
subject_n |
the number of subjects |
time_n |
the number of time points |
alpha |
FILL |
s1 |
FILL |
sim_seed |
the random seed with which to simulate the data |
a named list
X
Y
b
subject_ind
time_ind
Simulate data according to zero-inflated beta random effects model
simulate_zero_inflated_beta_random_effect_data( subject_n = 50, time_n = 5, v = 2, alpha = as.matrix(c(0, 0.5, -1)), beta = as.matrix(c(-0.5, -0.5, 0.5)), X = NA, Z = NA, s1 = 0.2, s2 = 0.2, sim_seed = 100 )simulate_zero_inflated_beta_random_effect_data( subject_n = 50, time_n = 5, v = 2, alpha = as.matrix(c(0, 0.5, -1)), beta = as.matrix(c(-0.5, -0.5, 0.5)), X = NA, Z = NA, s1 = 0.2, s2 = 0.2, sim_seed = 100 )
subject_n |
number of subjects |
time_n |
number of time points for each subject |
v |
the dispersion parameter in beta component |
alpha |
the coefficients in logistic component |
beta |
the coefficients in beta component |
X |
the covariates in logistic component |
Z |
the covariates in beta component |
s1 |
the stardard deviation of random effect in logistic component |
s2 |
the stardard deviation of random effect in beta component |
sim_seed |
the random seed |
a named list
Y the bacterial abundance generated from the model
X the covariates in logistic component
Z the covariates in beta component
alpha the coefficients in logistic component
beta the coefficients in beta component
s1 the stardard deviation of random effect in logistic component
s2 the stardard deviation of random effect in beta component
subject_ind the IDs for each subject
time_ind time points
simulate_zero_inflated_beta_random_effect_data( subject_n = 100, time_n = 5, X = as.matrix(c(rep(0, 50 * 5), rep(1, 50 * 5))), alpha = as.matrix(c(-0.5, 1)), beta = as.matrix(c(-0.5, 0.5)), s1 = 1, s2 = 0.8, v = 5, sim_seed = 100 )simulate_zero_inflated_beta_random_effect_data( subject_n = 100, time_n = 5, X = as.matrix(c(rep(0, 50 * 5), rep(1, 50 * 5))), alpha = as.matrix(c(-0.5, 1)), beta = as.matrix(c(-0.5, 0.5)), s1 = 1, s2 = 0.8, v = 5, sim_seed = 100 )
Fit zero-inflated beta regression with random effects
zibr( logistic_cov, beta_cov, Y, subject_ind, time_ind, component_wise_test = TRUE, quad_n = 30, verbose = FALSE )zibr( logistic_cov, beta_cov, Y, subject_ind, time_ind, component_wise_test = TRUE, quad_n = 30, verbose = FALSE )
logistic_cov |
the covariates in logistic component |
beta_cov |
the covariates in beta component |
Y |
the response variable in the regression model |
subject_ind |
the variable for subject IDs |
time_ind |
the variable for time points |
component_wise_test |
whether to perform component wise test. If true, ZIBR will calculate p-values for logistic and beta component respectively. |
quad_n |
Gaussian quadrature points |
verbose |
print the fitting process |
a named list
logistic_est_table - the estimated coefficients for logistic component.
logistic_s1_est - the estimated standard deviation for the random effect in the logistic component.
beta_est_table - the estimated coefficients for logistic component.
beta_s2_est - the estimated standard deviation for the random effect in the beta component.
beta_v_est - the estimated dispersion parameter in the beta component.
loglikelihood - the log likelihood of fitting ZIBR model on the data.
joint_p - the p-values for jointly testing each covariate in both logistic and beta component.
## simulate some data sim <- simulate_zero_inflated_beta_random_effect_data( subject_n = 100, time_n = 5, X = as.matrix(c(rep(0, 50 * 5), rep(1, 50 * 5))), Z = as.matrix(c(rep(0, 50 * 5), rep(1, 50 * 5))), alpha = as.matrix(c(-0.5, 1)), beta = as.matrix(c(-0.5, 0.5)), s1 = 1, s2 = 0.8, v = 5, sim_seed = 100 ) ## run zibr on the simulated data zibr_fit <- zibr( logistic_cov = sim$X, beta_cov = sim$Z, Y = sim$Y, subject_ind = sim$subject_ind, time_ind = sim$time_ind ) zibr_fit## simulate some data sim <- simulate_zero_inflated_beta_random_effect_data( subject_n = 100, time_n = 5, X = as.matrix(c(rep(0, 50 * 5), rep(1, 50 * 5))), Z = as.matrix(c(rep(0, 50 * 5), rep(1, 50 * 5))), alpha = as.matrix(c(-0.5, 1)), beta = as.matrix(c(-0.5, 0.5)), s1 = 1, s2 = 0.8, v = 5, sim_seed = 100 ) ## run zibr on the simulated data zibr_fit <- zibr( logistic_cov = sim$X, beta_cov = sim$Z, Y = sim$Y, subject_ind = sim$subject_ind, time_ind = sim$time_ind ) zibr_fit