stereo.algorithm.time_series_analysis.TimeSeriesAnalysis.main

TimeSeriesAnalysis.main(run_method='tvg_marker', use_col='timepoint', branch=None, p_val_combination='FDR', cluster_number=6, **kwargs)[source]
Parameters:

  • run_method – the model type when the algorithm is run, default = tvg_marker. tvg_marker: Calculate time variable gene based on expression of celltypes in branch other: Use fuzzy C means cluster method to cluster genes based on 1-p_value of celltypes in branch

  • use_col – the col in obs representing celltype or clustering

  • branch – celltypes order in use_col

  • p_val_combination – p_value combination method to use, choosing from [‘fisher’, ‘mean’, ‘FDR’], default to ‘FDR’, only for tvg_marker run_method

  • cluster_number – number of cluster, defaults to 6, only for other run_method

other parameters:

Note

All the parameters below are key word arguments and only for tvg_marker run_method.

Parameters:

  • statistical_test_method – one tail statistical test method to be used for up or down regulated examination between time point, choosing from [‘t-test’, ‘permutation’, ‘wilcoxon’], default to ‘t-test’]

  • permutation_batch – permutation_batch to control memory consumption if statistical_test_method == ‘permutation’, the higher permutation_batch require higher memory consumption. default to 100.

  • permutation_n_resamples – the times of resamples if statistical_test_method == ‘permutation’, default to 999.

Note

All the parameters below are key word arguments and only for other run_method.

Parameters:

  • spatial_weight – the weight to combine spatial feature, defaults to 1.

  • n_spatial_feature – n top features to combine of spatial feature, defaults to 2.

  • temporal_mean_threshold – filter out genes of which mean absolute temporal feature <= temporal_mean_threshold, defaults to 0.85.

  • temporal_top_threshold – filter out genes of which top absolute temporal feature < temporal_top_threshold, defaults to 1.

  • cluster_method – method to cluster gene based on spatial and temporal feature, choose from [‘fuzzy_C_means’, ‘gaussian_mixture’], default to fuzzy_C_means.

  • Epsilon – max value to finish iteration if cluster_method==’fuzzy_C_means’, defaults to 1e-7.

  • w_size – window size to rasterizing spatial expression, default to 20.

  • use_col – the col in obs representing celltype or clustering, default to None.

  • branch – celltypes order in use_col, default to None.

  • seed – fix seed in numpy to keep output constant in every run.