Release History

Version 0.1.82

updated for compatibilty with umap-learn >=0.5.0
bugfix of argument labels (predetermined cluster labels, overrides via’s clustering): edited the api to clarify that this needs to be of type and size ndarray (nsamples, )
updated function via_umap() to run with simply via_umap(via_object = v0) or via_umap(via_object = v0, init='via')

added lineage pathway visualization to improve the existing edge plotting function plot_edge_bundle()
for plot_edge_bundle, the parameter lineage_pathway:list = [] can be filled with cluster labels from the list of terminal cluster lineages in order to see the fine-grained lineage pathways along edges
example: plot_edge_bundle(via_object=v0, lineage_pathway=[7,10,9], linewidth_bundle=0.5, headwidth_bundle=2, cmap='plasma',text_labels=True, show_milestones=True, scale_scatter_size_pop=True)
if you wish to recompute the edge based visualization with different resolution you have two options:
1. set attribute via_object.hammerbundle_milestone_dict = None and then rereun plot_edge_bundle()
1. Or via_object.hammerbundle_milestone_dict=make_edgebundle_milestone(via_object=v0, n_milestones=40) followed by plot_edge_bundle(...)
Examine the single-cell lineage probabilities of each cell towards a particular terminal state using:
1. see via_object.single_cell_bp_rownormed
1. see via_object.single_cell_bp is actually lineage normalized so that the probabilities along a very rare lineage are not hidden.

bugfix draw_sc_lineage_probability() to allow plotting multiple and single lineages in same plot

improved draw_sc_lineage_probability() to correctly assign subplots when marker_lineages = [] is given by user as a subset of the terminal_clusters attribute of via. (corrections to the update in 0.1.57)
control fontsize
New Feature! VIA can autocompute via-umap and via-mds by passing do_compute_embedding = True and embedding_type = 'via-mds' or 'via-umap' when initializing a via object. e.g. v0 = via.VIA(X_data....) or after doing v0.run_via()

plot_gene_trend_heatmaps(via_object, df_gene_exp:pd.DataFrame,...)

New Feature! annotate nodes in plot_edge_bundle() by setting text_labels = True uses true_labels as annotations. Optionally provide list of single cell annotations length n_samples to use instead of true_labels. Example figure below

plot_edge_bundle(via_object, text_labels=True)

Bug fix for root detection (the initialization for None was over-writing RNA-velocity predicted roots)

corrected the auto-scaling in draw_sc_lineage_probability() so that each subplot has the same colorbar scale

optionally allow user to fix terminal states based on cell index or group label (corresponding to true_label)
optionally allow user to plot only selected lineages (by corresponding terminal cluster number) in get_gene_expression and draw_sc_lineage_probability (marker_lineages = [2,8,10])

support via-guided embeddings. In particular fast via-mds and via-umap which can be adjusted using known time-series data if available