omicverse.pp.qc

Contents

omicverse.pp.qc#

omicverse.pp.qc(adata, *, mode='seurat', min_cells=3, min_genes=200, nmads=5, max_cells_ratio=1, max_genes_ratio=1, batch_key=None, doublets=True, doublets_method='scdblfinder', filter_doublets=True, path_viz=None, tresh=None, mt_startswith='auto', mt_genes=None, ribo_startswith=('RPS', 'RPL'), ribo_genes=None, hb_startswith='^HB[^(P)]', hb_genes=None, use_gpu=True, batch_wise_mad=None, **kwargs)[source]#

Perform quality control on a dictionary of AnnData objects.

Parameters:

  • adata – AnnData object

  • mode (str (default: 'seurat')) – The filtering method to use. Valid options are ‘seurat’

  • 'seurat'. (and 'mads'. Default is)

  • min_cells (int (default: 3)) – The minimum number of cells for a sample to pass QC. Default is 3.

  • min_genes (int (default: 200)) – The minimum number of genes for a cell to pass QC. Default is 200.

  • max_cells_ratio (float (default: 1)) – The maximum number of cells ratio for a sample to pass QC. Default is 1.

  • max_genes_ratio (float (default: 1)) – The maximum number of genes ratio for a cell to pass QC. Default is 1.

  • nmads (int (default: 5)) – The number of MADs to use for MADs filtering. Default is 5.

  • doublets (bool (default: True)) – Whether to perform doublet detection. Default is True.

  • doublets_method (str (default: 'scdblfinder')) – The doublet detection method to use. Options are ‘scrublet’, ‘sccomposite’, ‘doubletfinder’, or ‘scdblfinder’. Default is ‘scdblfinder’ (Python port of R scDblFinder; xgboost on kNN+cxds features) for cpu and cpu-gpu-mixed modes, falling back to ‘scrublet’ if pyscdblfinder is not installed. The RAPIDS gpu mode keeps ‘scrublet’ as default for backwards compatibility.

  • filter_doublets (bool (default: True)) – Whether to filter out doublets (True) or just flag them (False). Default is True.

  • path_viz (default: None) – The path to save the QC plots. Default is None.

  • tresh (default: None) – A dictionary of QC thresholds. The keys should be ‘mito_perc’,

  • 'nUMIs' – Only used if mode is ‘seurat’. Default is None.

  • 'detected_genes'. (and) – Only used if mode is ‘seurat’. Default is None.

  • mt_startswith (str (default: 'auto')) – The prefix of mitochondrial genes. Default is ‘auto’, which automatically detects the prefix (e.g. ‘MT-’ for human, ‘mt-’ for mouse). Set explicitly (e.g. ‘MT-’) to override.

  • mt_genes (default: None) – The list of mitochondrial genes. Default is None.

  • None (if mt_genes is not)

  • ignored. (mt_startswith will be)

Returns:

An AnnData object containing cells that passed QC filters.

Return type:

adata

Examples

>>> import omicverse as ov
>>> adata = ov.pp.qc(adata, tresh={'mito_perc': 0.2, 'nUMIs': 500, 'detected_genes': 250})
>>> adata = ov.pp.qc(adata, mode='mads', nmads=5, doublets=True)
>>> # Auto-detects 'mt-' for mouse data
>>> adata = ov.pp.qc(adata)
>>> # Explicit prefix
>>> adata = ov.pp.qc(adata, mt_startswith='mt-')
Contents