AttributeError: module 'mofapy' has no attribute 'core.entry_point'
dfermin opened this issue · comments
Hi
This is a weird problem.
I've installed MOFA and I'm trying to work through the CLL example but can't execute the runMOFA(MOFAObject) command.
I keep getting this error:
x <- runMOFA(MOFAObject)
[1] "No output file provided, using a temporary file..."
Error in py_get_attr_impl(x, name, silent) :
AttributeError: module 'mofapy' has no attribute 'core.entry_point'
I'm using an Anaconda environment running Python 3.6.
It seems that reticulate has detected the correct instance and sees the mofapy package:
> py_config()
python: /home/dfermin/anaconda3/envs/py36/bin/python
libpython: /home/dfermin/anaconda3/envs/py36/lib/libpython3.6m.so
pythonhome: /home/dfermin/anaconda3/envs/py36:/home/dfermin/anaconda3/envs/py36
version: 3.6.7 | packaged by conda-forge | (default, Jul 2 2019, 02:18:42) [GCC 7.3.0]
numpy: /home/dfermin/anaconda3/envs/py36/lib/python3.6/site-packages/numpy
numpy_version: 1.17.0
mofapy: /home/dfermin/anaconda3/envs/py36/lib/python3.6/site-packages/mofapy
When I go to the site-packages directory I see mofapy and the entry_point.py script is there:
ls /home/dfermin/anaconda3/envs/py36/lib/python3.6/site-packages/mofapy/core/
BayesNet.py
build_model.py
distributions.py
entry_point.py
init_nodes.py
__init__.py
mixed_nodes.py
multiview_nodes.py
nodes.py
nongaussian_nodes.py
__pycache__
simulate.py
updates.py
utils.py
variational_nodes.py
I can import the library in python without any issues.
I'm not using Rstudio, this is all on the terminal on an instance of RHEL.
I've tried multiple versions of R and I'm currently using a freshly compiled version of 3.6.1.
Does anyone have a suggestion for a work around?
The R sessionInfo outptu is given below in case it helps.
> sessionInfo()
R version 3.6.1 (2019-07-05)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Red Hat Enterprise Linux Workstation 7.7 (Maipo)
Matrix products: default
BLAS/LAPACK: /home/dfermin/apps/openBLAS.libs/lib/libopenblas_haswellp-r0.3.6.so
locale:
[1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
[3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
[5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
[7] LC_PAPER=en_US.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] reticulate_1.13 MOFAdata_1.1.0 MOFA_1.1.1
loaded via a namespace (and not attached):
[1] SummarizedExperiment_1.14.1 beeswarm_0.2.3
[3] tidyselect_0.2.5 purrr_0.3.3
[5] reshape2_1.4.3 corrplot_0.84
[7] rhdf5_2.28.1 lattice_0.20-38
[9] colorspace_1.4-1 stats4_3.6.1
[11] rlang_0.4.1 pillar_1.4.2
[13] glue_1.3.1 BiocParallel_1.18.1
[15] BiocGenerics_0.30.0 RColorBrewer_1.1-2
[17] matrixStats_0.55.0 GenomeInfoDbData_1.2.1
[19] foreach_1.4.7 plyr_1.8.4
[21] stringr_1.4.0 zlibbioc_1.30.0
[23] munsell_0.5.0 gtable_0.3.0
[25] codetools_0.2-16 Biobase_2.44.0
[27] IRanges_2.18.3 doParallel_1.0.15
[29] MultiAssayExperiment_1.10.4 GenomeInfoDb_1.20.0
[31] vipor_0.4.5 parallel_3.6.1
[33] Rcpp_1.0.2 scales_1.0.0
[35] DelayedArray_0.10.0 S4Vectors_0.22.1
[37] jsonlite_1.6 XVector_0.24.0
[39] ggplot2_3.2.1 stringi_1.4.3
[41] dplyr_0.8.3 ggrepel_0.8.1
[43] cowplot_1.0.0 GenomicRanges_1.36.1
[45] grid_3.6.1 tools_3.6.1
[47] bitops_1.0-6 magrittr_1.5
[49] RCurl_1.95-4.12 lazyeval_0.2.2
[51] tibble_2.1.3 crayon_1.3.4
[53] pkgconfig_2.0.3 pheatmap_1.0.12
[55] Matrix_1.2-17 ggbeeswarm_0.6.0
[57] assertthat_0.2.1 iterators_1.0.12
[59] Rhdf5lib_1.6.3 R6_2.4.0
[61] compiler_3.6.1
I've been trying to debug this and found this inconsistency. I don't know if it will help
When I try to access the data in the mofa.core module from within R I cannot see any of the functions:
> library(reticulate)
> use_condaenv(condaenv="py36", required=TRUE)
> mofa <- import("mofapy")
> py_list_attributes(mofa$core)
[1] "__builtins__" "__cached__" "__doc__" "__file__"
[5] "__loader__" "__name__" "__package__" "__path__" "__spec__"
> dir("/home/dfermin/anaconda3/envs/py36/lib/python3.6/site-packages/mofapy/core")
[1] "__init__.py" "__pycache__" "BayesNet.py" "build_model.py"
[5] "distributions.py" "entry_point.py" "init_nodes.py" "mixed_nodes.py"
[8] "multiview_nodes.py" "nodes.py" "nongaussian_nodes.py" "simulate.py"
[12] "updates.py" "utils.py" "variational_nodes.py"
If I load a different python module the functions are visible and correspond to the python scripts for the module:
> vcf <- import("vcf") ## importing the PyVCF module
> py_list_attributes(vcf)
[1] "Filter" "RESERVED_FORMAT" "RESERVED_INFO" "Reader"
[5] "SampleFilter" "VCFReader" "VCFWriter" "VERSION"
[9] "Writer" "__builtins__" "__cached__" "__doc__"
[13] "__file__" "__loader__" "__name__" "__package__"
[17] "__path__" "__spec__" "filters" "model"
[21] "parser" "sample_filter"
>dir("home/dfermin/anaconda3/envs/py36/lib/python3.6/site-packages/vcf/")
[1] "__init__.py" "__pycache__" "cparse.pyx" "filters.py"
[5] "model.py" "parser.py" "sample_filter.py" "test"
[9] "utils.py"
If I switch over to python I have the same error that I observe in R:
$ conda activate py36
$ python
>>> from mofapy import core
>>> core
<module 'mofapy.core' from '/home/dfermin/anaconda3/envs/py36/lib/python3.6/site-packages/mofapy/core/__init__.py'>
>>> core.entry_point
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError: module 'mofapy.core' has no attribute 'entry_point'
This doesn't happen with the vcf module I tested earlier.
I don't know if this is because the vcf module doesn't contain a core subdirectory or not but I thought someone with a better understanding of python might be able to suggest a fix.
if somebody is still interested:
I had the same problem when installing mofapy directly in my conda-environment. After a long trial and error an installation from R helped to solve the problem, but I still do not know why:
run in R:
library(reticulate)
py_install("mofapy", envname = "mofa_env", method="conda",
conda = "/opt/anaconda/3-5.0.1/bin/conda", pip=TRUE)
additionally restart the R session afterwards
I had the same issue running MOFA2 on RStudio Server. The error message said also: "AttributeError: module 'mofapy' has no attribute 'core.entry_point'" I noticed that r-reticulate generated a conda env by itself.
So, I did not changed the conda_env for my MOFA run and installed mofapy2 by first accessing the r-reticulate env:
conda activate /home/XXX/.local/share/r-miniconda/envs/r-reticulate
And install mofapy2:
pip install mofapy2
That solved the problem for me. Hope it is helpful.
Been running into this as well today, and I found the following helpful to debug (using MOFA2 and use_basilisk=FALSE):
- Check
reticulate::py_config()which python installation is used - Open this python in your terminal and try to import the module
from mofapy2.run import entry_point
The error message you get will be more helpful than what run_mofa displays