Problems with figures generated by scv.pl.velocity_graph(adata)

[ETAOIK]

Discussed in #995

^{Originally posted by ETAOIK February 1, 2023}
Hi, I am new to scVelo application and I have some problems during studying the tutorial 'Velocity graph and pseudotime' in https://scvelo.readthedocs.io/en/stable/VelocityBasics/,
In my output, cell-to-cell transitions didn't overlap with the umap clusters plot like following(Fig.1),
scv.pl.velocity_graph(adata,threshold=0.2)

only set arrows=True could made it like the following(Fig.2), and in this situation edges_on_top=None didn't works

also, the the velocity graph generated by scv.pl.velocity_graph(adata) is quite small and the legend is quite big(Fig.3)
x, y = scv.utils.get_cell_transitions(adata, basis='umap', starting_cell=70)
ax = scv.pl.velocity_graph(adata, c='lightgrey', edge_width=.05, show=False)
ax = scv.pl.scatter(adata, x=x, y=y, s=120, c='ascending', cmap='gnuplot', ax=ax)

My question is how to deal with it. Any help is greatly appreciated!

[River366]

In fact, I had the same problem.

• [ Versions ]
  anndata 0.8.0
  scanpy 1.9.1
  PIL 9.4.0
  asttokens NA
  backcall 0.2.0
  beta_ufunc NA
  binom_ufunc NA
  cffi 1.15.1
  colorama 0.4.6
  comm 0.1.2
  cycler 0.10.0
  cython_runtime NA
  dateutil 2.8.2
  debugpy 1.6.6
  decorator 5.1.1
  executing 1.2.0
  h5py 3.8.0
  hypergeom_ufunc NA
  igraph 0.10.4
  invgauss_ufunc NA
  ipykernel 6.21.0
  jedi 0.18.2
  joblib 1.2.0
  kiwisolver 1.4.4
  leidenalg 0.9.1
  llvmlite 0.39.1
  louvain 0.8.0
  matplotlib 3.6.3
  mpl_toolkits NA
  natsort 8.2.0
  nbinom_ufunc NA
  ncf_ufunc NA
  nct_ufunc NA
  ncx2_ufunc NA
  numba 0.56.4
  numexpr 2.8.3
  numpy 1.23.5
  packaging 23.0
  pandas 1.5.3
  parso 0.8.3
  pexpect 4.8.0
  pickleshare 0.7.5
  pkg_resources NA
  platformdirs 2.6.2
  prompt_toolkit 3.0.36
  psutil 5.9.4
  ptyprocess 0.7.0
  pure_eval 0.2.2
  pydev_ipython NA
  pydevconsole NA
  pydevd 2.9.5
  pydevd_file_utils NA
  pydevd_plugins NA
  pydevd_tracing NA
  pygments 2.14.0
  pyparsing 3.0.9
  pytz 2022.7.1
  scipy 1.10.0
  scvelo 0.2.5
  session_info 1.0.0
  setuptools 66.1.1
  setuptools_scm NA
  six 1.16.0
  skewnorm_ufunc NA
  sklearn 1.2.1
  stack_data 0.6.2
  texttable 1.6.7
  threadpoolctl 3.1.0
  tornado 6.2
  traitlets 5.9.0
  typing_extensions NA
  wcwidth 0.2.6
  zipp NA
  zmq 25.0.0
  IPython 8.9.0
  jupyter_client 8.0.2
  jupyter_core 5.2.0
  Python 3.8.15 | packaged by conda-forge | (default, Jan 26 2023, 10:47:49) [GCC 11.3.0]
  Linux-5.15.0-58-generic-x86_64-with-glibc2.10

---

[ETAOIK]

I modified the source codes, and found one solution, although it didn't solve the problems fundamentally (cause if you didn't set arrows=True, the changes would not work), so here is my solution:
modify the codes of /scvelo-master/scvelo/plotting/velocity_graph like following (Line 83)

1. change the zorder from 0 to 2 (let the zorder bigger than that of edges) first,
2. and then add fontsize and figsize parameters statements (Set according to your needs):

Before changed:
ax = scatter(adata, layer=layer, color=color, size=size, ax=ax, zorder=0, **kwargs)
After changed:
ax = scatter(adata, layer=layer, color=color, size=size, ax=ax, zorder=2, fontsize=10,figsize=(8,6), **kwargs)

velocity graph and transition plot before changes:

velocity graph and transition plot after changes:

the following plot should add one more zorder statement (zorder=3)
x, y = scv.utils.get_cell_transitions(adata, basis='umap', starting_cell=70)
ax = modified_velocity_graph(adata, arrows=True, c='lightgrey', edge_width=.05, show=False)
ax = scv.pl.scatter(adata, x=x, y=y, s=120, c='ascending', cmap='gnuplot', ax=ax, zorder=3)

Although this solution could generate the correct figures, it only works in the situation of setting arrows=True, setting arrows=False will still lead to the wrong layer overlapping. And I would like to mention, setting arrows=True will cost more time (more computing power is cost in the process of computing directed graph?). I don't know how to correct it fundamentally, but at least now, we have the correct output figures. Hope it would be solved fundamentally asap. Thanks all

[joe-jhou2]

It works! Thanks.

[River366]

Dear author, thanks for your help!

[dburkhardt]

@adamgayoso @WeilerP just an FYI this is still an open issue. The is a workaround that involved patching the package, but it's incomplete (e.g. doesn't work with arrows=True)

[argearriojas]

The latest versions of matplotlib break the velocity_graph method. Downgrading to matplotlib==3.5.3 fixes this issue for me.

Until this compatibility issue is fixed, I would suggest adding version constraints to the requirements.txt file, or at least put a list of tested versions in the README file. This would also be relevant for versions of numpy and pandas that are currently incompatible with scvelo.

[WeilerP]

Fixed by #1162.