Getting segmentation fault(core dumped) error while converting anndata to SCE object across multiple function calls

[ankur26197-acog]

Hi,
I am analysing scRNA seq data using R for implementing all the steps and anndata2ri for interchangeability of anndata to SCE objects. I have wrapped these steps into separate functions. After running the first step i.e. QC, when I tried to execute the next step, I got segmentation fault(core dumped) while converting anndata to SCE object. Interestingly if I execute these functions in separate notebooks, it works.

I tried a couple of approaches like increasing memory for notebook, upgrading packages but nothing worked . At last when I passed the copy of anndata instead of the actual object, the issue was resolved. I am wondering what could be the reason for that.

Steps to reproduce-

1. Load a scRNA dataset in anndata.
2. Write functions for analysis like QC, normalisation, integration, clustering, etc. in R
3. Source each of these function using r from rpy2.robjects module and call them from a notebook/python file. Make sure all these functions calling the respective R functions are in the same notebook/python file. Pass the same anndata object modified after every function call.
4. After running QC, during the next function call while converting anndata to SCE, the kernel will die if running in a notebook else you will get a segmentation fault if running in a python file. Use faulthandler for a detailed stacktrace.

Ubuntu 20.04 Python version 3.8.8 anndata version 0.8.0 anndata2ri version 1.0.3 rpy version 3.2.7

Docker is used for complete setup.

[flying-sheep]

Hi, unfortunately those steps don’t help. You’ll need to be far more specific before we can help you.

Please check out this tutorial: http://sscce.org/