Repository navigation
[BUG] XeniumSample - No attribute embeddings_dict, missing packages installation #112
Description
Hello everyone,
I currently try to re-segment our Xenium data including a scRNAseq reference dataset. I came across a few errors which I managed to fix, but now I'm sadly stuck. I installed via pulling from github (Date: 04.06.25) :)
This is my code for importing
from segger.data.io import XeniumSample
from pathlib import Path
from segger.data.utils import calculate_gene_celltype_abundance_embedding
import scanpy as sc
import os
The point where I am stuck is here:
#Setup Xenium sample to create dataset
xs = XeniumSample(verbose=False, embedding_df=gene_celltype_abundance_embedding)
xs.set_file_paths(
transcripts_path=f'{raw_data_dir}/transcripts.parquet',
boundaries_path=f'{raw_data_dir}/nucleus_boundaries.parquet',
)
xs.set_metadata()
# Set the embedding to "cell_type_abundance" to use it in further processing
xs.set_embedding("cell_type_abundance")
And now I am stuck with the following error:
AttributeError Traceback (most recent call last)
Cell In[37], line 10
7 xs.set_metadata()
9 # Set the embedding to "cell_type_abundance" to use it in further processing
---> 10 xs.set_embedding("cell_type_abundance")
File ~/segger_dev/src/segger/data/io.py:452, in SpatialTranscriptomicsSample.set_embedding(self, embedding_name)
443 def set_embedding(self, embedding_name: str) -> None:
444 """
445 Set the current embedding type for the transcripts.
446
(...) 450
451 """
--> 452 if embedding_name in self.embeddings_dict:
453 self.current_embedding = embedding_name
454 else:
AttributeError: 'XeniumSample' object has no attribute 'embeddings_dict'
Could you please let me know if there is something wrong with my data or this is command is under construction currently? I would be really happy to use segger for improving our segmentation of the heart really soon. It would be great if there's a way how we could get this fixed :)
As a note: the code from the documentation did not work for me (you might be aware of this already). Also I recognized one needs to install torch and dask_geopandas before running the script :)
Thanks a lot and greets,
Lisa
Activity