omicverse.metabol.pyMetabo

class omicverse.metabol.pyMetabo(adata, random_state=0)

Lifecycle class for a metabolomics analysis.

Attributes populated as the pipeline runs

adata : the current state of the AnnData (updated in place on each call) raw : the original AnnData handed in at construction (never mutated) deg_table : DataFrame returned by differential() plsda_result : PLSDAResult from plsda() / opls_da()

param adata:

type adata:

AnnData

param random_state:

type random_state:

int (default: 0)

__init__(adata, random_state=0)

Parameters:

• adata (AnnData)

• random_state (int (default: 0))

Methods

__init__(adata[, random_state])
blank_filter(*, blank_mask[, ratio])	Drop features whose mean signal isn't ``ratio``× the blank mean.
cv_filter(*, qc_mask[, cv_threshold])	Drop features with QC coefficient-of-variation above cv_threshold.
differential(*[, group_col, group_a, ...])	Two-group univariate test across all metabolites; stores self.deg_table.
drift_correct(*, injection_order, qc_mask[, ...])	Correct injection-order drift via LOESS regression on QC samples.
impute(*[, method, seed])	Impute missing values in adata.X and return self.
normalize(*[, method])	Normalize each sample row to correct dilution and return self.
opls_da(*[, n_ortho, group_col, group_a, ...])	Fit OPLS-DA (one predictive + n_ortho orthogonal components).
plsda(*[, n_components, group_col, group_a, ...])	Fit PLS-DA and stash the result on self.plsda_result.
significant_metabolites(*[, padj_thresh, ...])	Filter self.deg_table to padj < padj_thresh and |log2fc| ≥ log2fc_thresh.
transform(*[, method])	Apply a feature-level transformation and return self.
vip_table()	Return VIP scores per metabolite — requires a prior PLS-DA / OPLS-DA fit.

Attributes

deg_table
plsda_result
random_state
adata
raw