pr_pseudotime result triming

[x1han]

Dear Palantir developers, @ManuSetty Palantir is an excellent tool for trajectory analysis, but I have a problem now. My input is a published dataset containing Oligodendrocytes lineage cells(Cell lines with precise differentiation characteristics, GSE75330), but the results obtained were not ideal. These are codes I used:

sc.pp.normalize_total(marques_ref_adata_sub_test, target_sum=1e4)
sc.pp.log1p(marques_ref_adata_sub_test)
sc.pp.highly_variable_genes(marques_ref_adata_sub_test)
sc.pp.pca(marques_ref_adata_sub_test)
# Run diffusion maps
pca_projections = pd.DataFrame(marques_ref_adata_sub_test.obsm['X_pca'], index=marques_ref_adata_sub_test.obs_names)
dm_res = palantir.utils.run_diffusion_maps(pca_projections, n_components=5)
ms_data = palantir.utils.determine_multiscale_space(dm_res)
sc.pp.neighbors(marques_ref_adata_sub_test)
sc.tl.umap(marques_ref_adata_sub_test)

start_cell = marques_ref_adata_sub_test[marques_ref_adata_sub_test.obs['cell_type'].isin(['OPC']), :].obs.index.tolist()[0]
terminal_states = pd.Series(['MOL2', 'MOL3'], index = [marques_ref_adata_sub_test[marques_ref_adata_sub_test.obs['cell_type'].isin(['MOL2']), :].obs.index.tolist()[0], marques_ref_adata_sub_test[marques_ref_adata_sub_test.obs['cell_type'].isin(['MOL3']), :].obs.index.tolist()[0]])

pr_res = palantir.core.run_palantir(ms_data, early_cell = start_cell, num_waypoints=500, terminal_states = terminal_states.index, use_early_cell_as_start = True)
pr_res.branch_probs.columns = terminal_states[pr_res.branch_probs.columns]
palantir.plot.plot_palantir_results(pr_res, umap)
marques_ref_adata_sub_test.obs['pr_pseudotime'] = pd.DataFrame(pr_res.pseudotime).reindex(marques_ref_adata_sub_test.obs_names)
scv.pl.scatter(marques_ref_adata_sub_test, basis='umap', color = ['pr_pseudotime', 'cell_type'], color_map='gnuplot2', legend_loc='on data')

and the results:

Sampling and flocking waypoints...
Time for determining waypoints: 0.004442989826202393 minutes
Determining pseudotime...
Shortest path distances using 30-nearest neighbor graph...
findfont: Font family ['Raleway'] not found. Falling back to DejaVu Sans.
findfont: Font family ['Raleway'] not found. Falling back to DejaVu Sans.
findfont: Font family ['Raleway'] not found. Falling back to DejaVu Sans.
findfont: Font family ['Raleway'] not found. Falling back to DejaVu Sans.
findfont: Font family ['Raleway'] not found. Falling back to DejaVu Sans.
findfont: Font family ['Raleway'] not found. Falling back to DejaVu Sans.
findfont: Font family ['Raleway'] not found. Falling back to DejaVu Sans.
findfont: Font family ['Raleway'] not found. Falling back to DejaVu Sans.
findfont: Font family ['Raleway'] not found. Falling back to DejaVu Sans.
findfont: Font family ['Raleway'] not found. Falling back to DejaVu Sans.
findfont: Font family ['Raleway'] not found. Falling back to DejaVu Sans.
Time for shortest paths: 0.8781302213668823 minutes
Iteratively refining the pseudotime...
Correlation at iteration 1: 0.9999
Correlation at iteration 2: 1.0000
Entropy and branch probabilities...
Markov chain construction...
Computing fundamental matrix and absorption probabilities...
Project results to all cells...

It seems that Palantir's results appear to have misidentified the NFOL2 cluster as a terminally differentiated state, although this was not actually the case, meanwhile, the result of scanpy.tl.dpt may be the right one.

I have to say that Palantir is quite an excellent tool and I am quite fond of it. Therefore, in this context, I have two concerns.

1. How can I remove the warning 'findfont: Font family ['Raleway'] not found. Falling back to DejaVu Sans.', (This occurred an excessive number of times. )
2. Is there any parameters that can adjust the pseudotime result to the right one: OPC-COP-NFOL-MFOL-MOL. (Given that the dpt results appear to align with the actual outcomes. )

Hoping for your reply!