multivelovae.differential_dynamics
- multivelovae.differential_dynamics(adata, adata_atac, model, group1=None, group2=None, group_key=None, idx1=None, idx2=None, batch_key=None, batch_correction=True, weight_batch_uniform=False, mode='vanilla', signed_velocity=True, test_decoupling=False, save_raw=False, n_samples=5000, norm=0.1, delta=1, fdr=0.05, seed=0)
Calculate differential dynamics between two groups of cells.
- Args:
- adata (
anndata.AnnData): RNA AnnData object.
- adata_atac (
anndata.AnnData): ATAC AnnData object.
- model (
velovae.VAEChrom): A trained MultiVeloVAE model.
- group1 (str, optional):
Name of group 1 in adata.obs[group_key]. Defaults to None.
- group2 (str, optional):
Name of group 2 in adata.obs[group_key]. Defaults to None.
- group_key (str, optional):
Field in adata.obs to find group labels. Defaults to None.
- idx1 ([list,
numpy.ndarray], optional): Indices of cells in group 1. Defaults to None.
- idx2 ([list,
numpy.ndarray], optional): Indices of cells in group 2. Defaults to None.
- batch_key (str, optional):
Field in adata.obs to find batch labels. Defaults to None.
- batch_correction (bool, optional):
Whether to perform batch correction before group comparisons. Defaults to True.
- weight_batch_uniform (bool, optional):
Whether to draw cells equally or based on cell counts in each batch. Defaults to False.
- mode (str, optional):
vanilla mode is one-sided and change mode is two-sided. change mode uses delta to specify null hypothesis and computes a P-value. Defaults to ‘vanilla’.
- signed_velocity (bool, optional):
Whether to use original velocity values or absolute values. Defaults to True.
- test_decoupling (bool, optional):
Whether to test decoupling and coupling dynamics. Defaults to False.
- save_raw (bool, optional):
Whether to save the predicted data in adata. Defaults to False.
- n_samples (int, optional):
Number of data points in each group to generate. Defaults to 5000.
- norm (Float, optional):
Normalization factor for log2 difference. Defaults to 0.1.
- delta (Float, optional):
Interval of change used to define null hypothesis. Defaults to 1.
- fdr (Float, optional):
False Discovery Rate threshold. Defaults to 0.05.
- seed (int, optional):
Seed for random generator. Defaults to 0.
- adata (
- Returns:
pandas.DataFrame: Output DataFrame with differential dynamics on various variables.