LikelihoodMethod.cluster_truth

LikelihoodMethod.cluster_truth#

missionbio.demultiplex.dna.likelihood.LikelihoodMethod.cluster_truth

static LikelihoodMethod.cluster_truth(dna: Assay, labels: ndarray, genome: Optional[str] = None) → Truth#

Variants that could be different across samples

This method assumes that the DNA assay has already been assigned labels and these are stored in the DEMUX_SAMPLE row attribute.

Parameters:

  • dna – The DNA assay in which the FILTERED column attribute is present. All variants passing filters will be used to generate the truth.

  • labels – The labels assigned to the cells. The truth will be created for each unique value in the labels excluding MIXED and AMBIGUOUS.

  • genome – The genome version used to fetch annotations e.g. “hg19”, “hg38”. If None, the “genome_version” metadata of the dna assay is used.

Returns:

The dataframe with extra variants

Raises:

  • ImputationError – When only one sample found or no cells labeled.

  • AnnotationAPIError – When the missionbio annotation API fails.

  • AllVariantsFilteredError – When all the variants get filtered out. This can happen if the variants are not differentiating between samples or the only variants available might be pathogenic.

< Class LikelihoodMethod

Contents