Perform Partial Least Squares Regression (PLSR) or sparse PLSR (sPLSR) analysis

This function wraps the mixOmics pls() and spls() functions to perform regression of one or more response variables on a set of predictors. It automatically handles optional predictor scaling, extraction of numeric columns and grouping information, and returns a small list of recorded plots for display in Shiny. The first plot (indiv_plot) is a score plot of the first two latent components colored by a grouping variable (if provided). The second plot (pred_plot) is a scatterplot of the observed vs predicted response values for the first response variable. Additional plots could easily be added (e.g. loadings), but these two provide basic diagnostics.

Usage

cyt_plsr(
  data,
  response_col,
  predictor_cols = NULL,
  group_col = NULL,
  ind_names = FALSE,
  comp_num = 2,
  sparse = FALSE,
  var_num = NULL,
  cv_opt = NULL,
  fold_num = 5,
  scale = NULL,
  custom_fn = NULL,
  ellipse = FALSE,
  pls_colors = NULL,
  output_file = NULL,
  font_settings = NULL,
  progress = NULL
)

Arguments

data: A data frame containing the predictor and response variables.
response_col: The name of the column containing the response variable. Must be numeric.
predictor_cols: Character vector of column names to use as predictors.
group_col: The name of the column to use for grouping samples in the score plot. If NULL, no grouping is applied.
ind_names: Logical, or a character vector. If TRUE, sample names are displayed on the plot. If a character vector, these names are used.
comp_num: The number of components to use in the PLSR model.
sparse: Logical. If TRUE, performs sparse PLSR (sPLSR).
var_num: The number of variables to keep per component for sPLSR. Required if sparse = TRUE.
cv_opt: Character string specifying the cross-validation option: "None", "LOOCV" (Leave-One-Out Cross-Validation), or "Mfold" (M-fold Cross-Validation).
fold_num: The number of folds for M-fold cross-validation. Only applicable if cv_opt = "Mfold".
scale: Character string specifying an optional transformation for numeric predictor columns. Supported values are NULL (default; no transformation), "none", "log2", "log10", "zscore", or "custom".
custom_fn: Optional transformation function used when scale = "custom".
ellipse: Logical. If TRUE, draws 95% confidence ellipses on the score plot.
pls_colors: A character vector of colors to use for grouping.
output_file: Optional. A file path to save the plots as a PDF.
font_settings: Optional named list of font sizes for supported plot text elements.
progress: Optional. A Shiny Progress object for reporting progress updates.

Value

A list containing:

scores_plot: A recorded plot of the PLSR scores.
pred_vs_obs: A recorded plot of predicted vs. observed response values.
residuals_plot: A recorded plot of residuals vs. fitted values.
cv_plot: A recorded plot of cross-validation performance (Q2 and RMSEP), if cv_opt is not "None".
loadings: A list of recorded plots for loadings of each component.
vip_scores: A list of ggplot objects of VIP scores for each component.
vip_scores_indiv: A recorded plot of the PLSR scores for variables with VIP > 1, if applicable.
vip_cv_plot: A recorded plot of cross-validation performance for variables with VIP > 1, if applicable.
metrics_text: Text summary of Q2 values, if available.
pdf_file: The path to the generated PDF file, if output_file was provided.

Author

Shubh Saraswat