Open KforKuma opened 10 months ago
KforKuma
commented
10 months ago Here comes another kind of bug: after I ran scv.tl.recover_dynamics(adata, n_jobs=8) and then gc.collect() it corrupted. Maybe I'd better make it a issue in scvelo too?
Marius1311
commented
10 months ago Here comes another kind of bug: after I ran
scv.tl.recover_dynamics(adata, n_jobs=8)and thengc.collect()it corrupted. Maybe I'd better make it a issue in scvelo too?
yes, that seems like an scvelo bug.
Marius1311
commented
10 months ago Hi @michalk8, do you have any idea of what could have caused the original bug here?
Marius1311
commented
6 months ago Does this issue still persist @KforKuma ?
KforKuma
commented
6 months ago nope... I tried a few times on several different python version & cellrank version. Somehow I believe this may be related to scikit-learn? Because I repetitively encounter It seems that scikit-learn has not been built correctly. sort of problem If I import cellrank (or any other package depends on sklearn) first , even though it seems to be installed nicely, as when I run python -c "import sklearn; sklearn.show_versions()" it returns:
System:
python: 3.10.14 | packaged by conda-forge | (main, Mar 20 2024, 12:45:18) [GCC 12.3.0]
executable: /data/HeLab/miniconda3/envs/cellranknew/bin/python
machine: Linux-2.6.32-431.11.2.el6.x86_64-x86_64-with-glibc2.17
Python dependencies:
sklearn: 1.4.2
pip: 24.0
setuptools: 69.5.1
numpy: 1.26.4
scipy: 1.13.0
Cython: None
pandas: 2.2.2
matplotlib: 3.8.4
joblib: 1.4.2
threadpoolctl: 3.5.0
Built with OpenMP: True
threadpoolctl info:
user_api: openmp
internal_api: openmp
num_threads: 24
prefix: libgomp
filepath: /data/HeLab/miniconda3/envs/cellranknew/lib/python3.10/site-packages/scikit_learn.libs/libgomp-a34b3233.so.1.0.0
version: None
user_api: blas
internal_api: openblas
num_threads: 24
prefix: libopenblas
filepath: /data/HeLab/miniconda3/envs/cellranknew/lib/libopenblasp-r0.3.27.so
version: 0.3.27
threading_layer: pthreads
architecture: SandybridgeI have to import sklearn firstly to avoid abovementioned error. Could this implies some problem there?
I followed up the tutorial here and with the pancreas exmaple dataset. Every time I went to the line
vk.compute_transition_matrix(), python produced a segfault. Sometimes it just corrupted, other times it printed very long output, which is attached below.Versions:
Besides my python is 3.10 and I am on a Centos 6 server. I'm a little bit concerned about using an old version of system, does it matter? I have seen many example when ppl using Ubuntu they simply sudo install libmalloc-minimal4 but it seems to be a different situation.