This is is the main function in v0 of verywise.
It checks the user inputs and runs the single_lmm function across all vertices.
Usage
run_vw_lmm(
formula,
pheno,
subj_dir,
outp_dir = NULL,
hemi = c("both", "lh", "rh"),
seed = 3108,
n_cores = 1,
chunk_size = 1000,
FS_HOME = Sys.getenv("FREESURFER_HOME"),
folder_id = "folder_id",
verbose = TRUE,
...
)Arguments
- formula
: Model formula object. This should specify a linear mixed model using `lme4` syntax. #' e.g.
vw_thickness ~ sex * age + site + (1|id). The *outcome* should be a brain surface metric. Available options: * `vw_thickness` * `vw_area` * `vw_area.pial` * `vw_curv` * `vw_jacobian_white` * `vw_pial` * `vw_pial_lgi` * `vw_sulc` * `vw_volume` * `vw_w_g.pct` * `vw_white.H` * `vw_white.K`- pheno
: the phenotype data object (already loaded in the global environment) or a string containing a file path. Supported file extensions are: rds, csv, txt and sav.
- subj_dir
: path to the FreeSurfer data, this expects a verywise structure.
- outp_dir
: output path, where do you want results to be stored. If none is provided by the user, a "results" sub-directory will be created inside
subj_dir.- hemi
: (default = "both") which hemispheres to run.
- seed
: (default = 3108) random seed.
- n_cores
: (default = 1) number of cores for parallel processing.
- chunk_size
: (default = 1000) size of data chunk for parallel processing
- FS_HOME
: FreeSurfer directory, i.e.
$FREESURFER_HOME.- folder_id
: (default = "folder_id") the name of the column in pheno that contains the directory names of the input neuroimaging data (e.g. "sub-10_ses-T1").
- verbose
: (default = TRUE)
- ...
Arguments passed on to
hemi_vw_lmmfwhm: (default = 10) full-width half maximum value
fs_template: (default = "fsaverage") template on which to register vertex-wise data. The following values are accepted: * fsaverage (default) = 163842 vertices (highest resolution), * fsaverage6 = 40962 vertices, * fsaverage5 = 10242 vertices, * fsaverage4 = 2562 vertices, * fsaverage3 = 642 vertices Note that, at the moment, these are only used to downsample the brain map, for faster model tuning. `verywise` expects the input data to be always registered on the "fsaverage" template and the final analyses should also be run using `fs_template = "fsaverage"` to avoid (small) imprecisions in vertex registration and smoothing.
apply_cortical_mask: (default = TRUE) remove vertices that are not on the cortex.
save_ss: (default = TRUE) save the super-subject matrix as an rds file for quicker processing in the future.
model: (default =
"lme4::lmer") # "stats::lm"use_model_template: pre-compile the model?