Step: Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA)

Perform OPLS-DA using glystats::gly_oplsda() and plot it with glyvis::plot_oplsda(). OPLS-DA separates variation into predictive (related to group) and orthogonal (unrelated) components. This step only works with binary classification (exactly 2 groups).

This step depends on the on parameter (default: exp).

• When on = "exp", requires exp (usually after step_preprocess()).

• When on = "sig_exp", requires sig_exp from one of step_dea_limma(), step_dea_ttest(), step_dea_wilcox(), step_dea_anova(), or step_dea_kruskal().

• When on = "trait_exp", requires trait_exp from step_derive_traits().

• When on = "sig_trait_exp", requires sig_trait_exp from DEA on traits.

• When on = "dynamic_motif_exp", requires dynamic_motif_exp from step_quantify_dynamic_motifs().

• When on = "sig_dynamic_motif_exp", requires sig_dynamic_motif_exp from DEA on motifs.

• When on = "branch_motif_exp", requires branch_motif_exp from step_quantify_branch_motifs().

• When on = "sig_branch_motif_exp", requires sig_branch_motif_exp from DEA on motifs.

Usage

step_oplsda(
  on = "exp",
  pred_i = 1,
  ortho_i = NA,
  scale = TRUE,
  plot_width = 5,
  plot_height = 5,
  ...
)

Arguments

on

Name of the experiment to run OPLS-DA on. Can be "exp", "sig_exp", "trait_exp", "sig_trait_exp", "dynamic_motif_exp", "sig_dynamic_motif_exp", "branch_motif_exp", "sig_branch_motif_exp".

pred_i

Number of predictive components to include. Default is 1.

ortho_i

Number of orthogonal components to include. Default is NA (automatic).

scale

Logical indicating whether to scale the data. Default is TRUE.

plot_width

Width of plots in inches. Default is 5.

plot_height

Height of plots in inches. Default is 5.

...

Additional arguments passed to glystats::gly_oplsda().

Value

A glysmith_step object.

Details

Data required:

• exp (if on = "exp"): The experiment to run OPLS-DA on

• trait_exp (if on = "trait_exp"): The trait experiment to run OPLS-DA on

• dynamic_motif_exp (if on = "dynamic_motif_exp"): The dynamic motif experiment to run OPLS-DA on

• branch_motif_exp (if on = "branch_motif_exp"): The branch motif experiment to run OPLS-DA on

Tables generated (with suffixes):

• oplsda_samples: A table containing the OPLS-DA scores for each sample

• oplsda_variables: A table containing the OPLS-DA loadings for each variable

• oplsda_variance: A table containing the explained variance for each component

• oplsda_vip: A table containing the Variable Importance in Projection (VIP) scores

• oplsda_perm_test: A table containing permutation test results

Plots generated (with suffixes):

• oplsda_scores: An OPLS-DA score plot colored by group

• oplsda_loadings: An OPLS-DA loading plot

• oplsda_variance: An OPLS-DA variance (scree) plot

• oplsda_vip: An OPLS-DA VIP score plot

AI Prompt

This section is for AI in inquire_blueprint() only.

• Include this step when users explicitly asks for OPLS-DA.

• This step only works with binary classification (exactly 2 groups). If multiple groups are found, ask if step_subset_groups() should be run first.

See also

glystats::gly_oplsda(), glyvis::plot_oplsda()

Examples

step_oplsda()
#> <step "step_oplsda()"> Orthogonal partial least squares discriminant analysis
step_oplsda(pred_i = 1, ortho_i = 1)
#> <step "step_oplsda(pred_i = 1, ortho_i = 1)"> Orthogonal partial least squares
#> discriminant analysis