omicverse.micro.DA.ancombc

Contents

omicverse.micro.DA.ancombc#

DA.ancombc(group_key, rank=None, min_prevalence=0.1, pseudocount=1.0)[source]#

ANCOM-BC via skbio.stats.composition.ancombc (skbio ≥ 0.7.1).

Parameters:

  • pseudocount (float (default: 1.0)) – Added to every cell before the test so the matrix has no zeros — scikit-bio’s ANCOM-BC explicitly rejects tables with zero or negative components. 1.0 (the default) is the classic count pseudocount; set to None / 0 to skip the replacement (e.g. if you’ve already applied skbio.stats.composition.multi_replace yourself).

  • note:: (..) – The scikit-bio ANCOM-BC API is evolving (the 0.7.1 return type is a named-tuple-like bundle, not a DataFrame). This wrapper expects the 0.7.1 shape with attributes lfc / se / W / p / q / diff_abn indexed by feature and will raise NotImplementedError on other shapes so the caller knows to pin scikit-bio.

  • group_key (str)

  • rank (Optional[str] (default: None))

  • min_prevalence (float (default: 0.1))

Contents