fit_nullmodel_genome_cond_spa: Fitting conditional generalized linear mixed models with...

In xihaoli/STAARpipelineSummary: Summarization and Visualization of Analysis Results Generated by STAARpipeline

Fitting conditional generalized linear mixed models with known relationship matrices for conditional analysis in imbalanced case-control setting.

Description

The fit_nullmodel_genome_cond_spa function fit regression models for conditional analysis in imbalanced case-control setting, which provides the preliminary step for subsequent conditional variant-set tests in conditional analysis. Each chromosome has a separate null model for conditional analysis. See fit_nullmodel for more details.

Usage

fit_nullmodel_genome_cond_spa(
  fixed,
  data = parent.frame(),
  kins,
  use_sparse = TRUE,
  use_SPA = TRUE,
  agds_dir,
  known_loci,
  geno_missing_imputation = c("mean", "minor"),
  MAC_cutoff = 20,
  output_path,
  cond_null_model_name = NULL,
  phenotype_id,
  phenotype,
  kins_cutoff = 0.022,
  id,
  random.slope = NULL,
  groups = NULL,
  family = binomial(link = "logit"),
  method = "REML",
  method.optim = "AI",
  maxiter = 500,
  tol = 1e-05,
  taumin = 1e-05,
  taumax = 1e+05,
  tauregion = 10,
  verbose = FALSE,
  ...
)

Arguments

fixed	an object of class formula (or one that can be coerced to that class): a symbolic description of the fixed effects model to be fitted. For multiple phenotype analysis, formula recognized by lm, such as cbind(y1,y2,y3) ~ x1 + x2, can be used in fixed as fixed effects.
data	a data frame or list (or object coercible by as.data.frame to a data frame) containing the variables in the model.
kins	a known positive semi-definite relationship matrix (e.g. kinship matrix in genetic association studies) or a list of known positive semi-definite relationship matrices. The rownames and colnames of these matrices must at least include all samples as specified in the id column of the data frame data. If kins is NULL, fit_nullmodel will switch to the generalized linear model with no random effects.
use_sparse	a logical switch of whether the provided dense kins matrix should be transformed to a sparse matrix (default = TRUE).
use_SPA	a logical switch determines if the null model fitting occurs in an imbalanced case-control setting (default = TRUE).
agds_dir	file directory of annotated GDS (aGDS) files for all chromosomes (1-22)
known_loci	the data frame of variants to be adjusted for in conditional analysis and should contain 4 columns in the following order: chromosome (CHR), position (POS), reference allele (REF), and alternative allele (ALT) (default = NULL).
geno_missing_imputation	method of handling missing genotypes. Either "mean" or "minor" (default = "mean").
MAC_cutoff	the cutoff of the minimum minor allele count of known variants adjusted in conditional analysis (default = 20).
output_path	the directory for the output files.
cond_null_model_name	the file name of conditional null models (default = NULL).
phenotype_id	id of samples.
phenotype	outcome in regression.
kins_cutoff	the cutoff value for clustering samples to make the output matrix sparse block-diagonal (default = 0.022).
id	a column in the data frame data, indicating the id of samples. When there are duplicates in id, the data is assumed to be longitudinal with repeated measures.
random.slope	an optional column indicating the random slope for time effect used in a mixed effects model for longitudinal data. It must be included in the names of data. There must be duplicates in id and method.optim must be "AI" (default = NULL).
groups	an optional categorical variable indicating the groups used in a heteroscedastic linear mixed model (allowing residual variances in different groups to be different). This variable must be included in the names of data, and family must be "gaussian" and method.optim must be "AI" (default = NULL).
family	a description of the error distribution and link function to be used in the model. This can be a character string naming a family function, a family function or the result of a call to a family function. (See family for details of family functions).
method	method of fitting the generalized linear mixed model. Either "REML" or "ML" (default = "REML").
method.optim	optimization method of fitting the generalized linear mixed model. Either "AI", "Brent" or "Nelder-Mead" (default = "AI").
maxiter	a positive integer specifying the maximum number of iterations when fitting the generalized linear mixed model (default = 500).
tol	a positive number specifying tolerance, the difference threshold for parameter estimates below which iterations should be stopped (default = 1e-5).
taumin	the lower bound of search space for the variance component parameter \tau (default = 1e-5), used when method.optim = "Brent". See Details.
taumax	the upper bound of search space for the variance component parameter \tau (default = 1e5), used when method.optim = "Brent". See Details.
tauregion	the number of search intervals for the REML or ML estimate of the variance component parameter \tau (default = 10), used when method.optim = "Brent". See Details.
verbose	a logical switch for printing detailed information (parameter estimates in each iteration) for testing and debugging purpose (default = FALSE).
...	additional arguments that could be passed to glm.

Value

The function returns objects of the null models fit from fit_nullmodel and whether the kins matrix is sparse when fitting the null model, each chromosome has one output. See fit_nullmodel for more details.

References

Chen, H., et al. (2016). Control for population structure and relatedness for binary traits in genetic association studies via logistic mixed models. The American Journal of Human Genetics, 98(4), 653-666. (pub)

Chen, H., et al. (2019). Efficient variant set mixed model association tests for continuous and binary traits in large-scale whole-genome sequencing studies. The American Journal of Human Genetics, 104(2), 260-274. (pub)

Chen, H. (2021). GMMAT: Generalized linear Mixed Model Association Tests Version 1.3.2. (web)

xihaoli/STAARpipelineSummary documentation built on July 27, 2024, 4:30 p.m.