Step: Preprocessing

Preprocess the experiment using glyclean::auto_clean(). This step can be omitted if the experiment is already preprocessed.

Usage

step_preprocess(
  batch_col = "batch",
  qc_name = "QC",
  normalize_to_try = NULL,
  impute_to_try = NULL,
  remove_preset = "discovery",
  batch_prop_threshold = 0.3,
  check_batch_confounding = TRUE,
  batch_confounding_threshold = 0.4,
  rep_col = NULL
)

Arguments

batch_col: Column name for batch information (for QC plots and batch effect handling).
qc_name: Name of QC sample group (used for QC sample detection in preprocessing).
normalize_to_try: Normalization methods to try during auto_clean.
impute_to_try: Imputation methods to try during auto_clean.
remove_preset: Preset for data removal: "discovery", "biomarker", or NULL.
batch_prop_threshold: Threshold for batch proportion filtering.
check_batch_confounding: Whether to check for batch confounding.
batch_confounding_threshold: Threshold for batch confounding detection.
rep_col: Column name for replicate information (for QC plots).

Value

A glysmith_step object.

Details

Data required:

exp: The experiment to preprocess

Data generated:

raw_exp: The raw experiment (previous exp, saved for reference)

This step is special in that it silently overwrites the exp data with the preprocessed experiment. This ensures that no matter if preprocessing is performed or not, the "active" experiment is always under the key exp. The previous exp is saved as raw_exp for reference.

AI Prompt

This section is for AI in inquire_blueprint() only.

Always include this step by default unless the user explicitly excludes it or tell you she/he has already performed preprocessing.
Ask for the column name for batch information if not provided.
Ask for QC samples in the experiment if not provided. If so, ask the group name of the QC samples. Explain to the user that if it is "QC" for example, the samples with "QC" in the group_col column will be considered as QC samples. And these QC samples will be used for choosing the best normalization and imputation methods. Also mention that QC samples will be excluded after preprocessing.
If the user intents to perform biomarker related analysis, set remove_preset to "biomarker".
Use default values for other arguments unless the user explicitly specifies otherwise.

Examples