Inquiry about Using Dynamo without unspliced/spliced Matrix

[linWiky]

Dear Dynamo Development Team,

Greetings!
First and foremost, I would like to express my heartfelt gratitude for your remarkable contribution to single-cell analysis with Dynamo. It is an impressive tool that has greatly advanced dynamic single-cell data analysis and pseudotime trajectory studies.

While studying the tutorial "Dynamo without unspliced/spliced matrix" (https://dynamo-release-test.readthedocs.io/en/latest/tutorials/notebooks/dynamo_beyondvelo.html), I noticed that Dynamo supports analysis without the upstream RNA velocity loom files. I attempted to use my single-cell raw data (triplicate files without unspliced/spliced information) to perform quality control, dimensional reduction, clustering, and pseudotime analysis using VIA. However, I encountered errors when executing the following steps provided in the tutorial:

I started by running dyn.pl.topography(adata, basis='umap', background='white', color=['ntr', 'clusters'], streamline_color='black', show_legend='on data', frontier=True).
Then, I proceeded with dyn.pp.recipe_monocle(adata, min_expr_cells=1, min_expr_avg=0, max_expr_avg=10, svr_gamma=None), followed by dyn.tl.reduceDimension(adata).
Next, I attempted to calculate cell velocities using dyn.tl.cell_velocities(adata, method='pearson', other_kernels_dict={'transform': 'sqrt'}) and ran into errors.
Afterward, I tried additional vector field-related analyses such as dyn.vf.VectorField(adata, basis='pca', M=1000, pot_curl_div=True), dyn.vf.speed(adata, basis='pca'), dyn.vf.curl(adata, basis='umap'), dyn.vf.divergence(adata, basis='pca'), dyn.vf.acceleration(adata, basis='pca'), and dyn.vf.curvature(adata, basis='pca'), but encountered further issues.
The problem specifically arises during the dyn.tl.cell_velocities() step and subsequent vector field-related analyses, where errors prevent the expected results from being generated.

I would like to inquire about the following:

Does Dynamo inherently not support vector field analyses in the absence of unspliced/spliced matrices?
If such analyses are supported, are there specific adjustments needed for the input data or the analysis workflow?
I would greatly appreciate your guidance on this matter. Thank you once again for your outstanding work!

Best regards,
Linyunhua

[Sichao25]

Hi, thanks for raising this discussion. Dynamo does support vector field analysis in the absence of unspliced/spliced matrices. Based on the information you provided, it seems you may have missed the step of converting pseudotime to velocities through dyn.tl.pseudotime_velocity. (Please note that dyn.tl.cell_velocities is used for mapping velocities to reduced dimensions, usually after estimating velocities with dyn.tl.dynamics.) Let me know if this helps!

[linWiky]

Thank you for your reply. I have already run the seudotime to velocities conversion process you mentioned, but still get an error. I have attached the analysis error message and data frame. I look forward to your early reply.

[linWiky]