Error with gene_covariation_analysis function

[reinertanalytics]

Hello! I am trying to use the gene_covariation_analysis for ligand receptor analysis and am running into an error. Here is the function as I have it written:

cov_out=scov.gene_covariation_analysis(iNMFmode=True,
Radius=inputRadius,
no_of_factors=3,
refpath=ref_datapath,
quepath=query_datapath,
spatial_integration_modality='double',
output_niche_prediction_dir=output_nico_dir,
ref_cluster_tag=ref_cluster_tag,
LRdbFilename='/Users/PATHNiCoLRdb.txt'
)

and here is the error I am getting:

---

AttributeError Traceback (most recent call last)
Cell In[38], line 9

----> 9 cov_out=scov.gene_covariation_analysis(iNMFmode=True,
10 Radius=inputRadius,
11 no_of_factors=3,
12 refpath=ref_datapath,
13 quepath=query_datapath,
14 spatial_integration_modality='double',
15 output_niche_prediction_dir=output_nico_dir,
16 ref_cluster_tag=ref_cluster_tag,
17 LRdbFilename='/Users/PATH/NiCoLRdb.txt'
18 )

File /opt/homebrew/Caskroom/miniconda/base/envs/nico_env/lib/python3.11/site-packages/nico/Covariations.py:274, in gene_covariation_analysis(Radius, output_niche_prediction_dir, refpath, quepath, ref_cluster_tag, ref_original_counts, LRdbFilename, iNMFmode, no_of_factors, shap_analysis, shap_cluster_cutoff, cutoff_to_count_exp_cell_population, seed, spatial_integration_modality, anndata_object_name, lambda_c, coeff_cutoff_for_rid_reg, logistic_coef_cutoff)
265 gene_set_names=[]
267 #print('sc1 annotation_singlecell_cluster_id',len(annotation_singlecell_cluster_id))
268 #print('sc2 annotation_singlecell_barcode_id',len(annotation_singlecell_barcode_id))
269 #print('sc3 annotation_singlecell_celltypename',len(annotation_singlecell_celltypename))
(...)
272
273 # load spatial dat
--> 274 sp_genename=sct_ad_sp.var_names.to_numpy()
275 sc_genename=sct_ad_sc.var_names.to_numpy()
276 index_sp,index_sc=find_index(sp_genename,sc_genename)

AttributeError: 'NoneType' object has no attribute 'var_names'

I have checked and my spatial data does have var names:
Gene names (.var_names): Index(['ACSM2B', 'ACTA2', 'ACTG2', 'ADAMTS1', 'ADAMTS6', 'ADGRE2', 'ADGRL3',
'AGTR1', 'ALDH1A2', 'ANO3'],
dtype='object').

I have checked that my gene names are strings. My spatial data is called "sct_spatial.h5ad" and the ref is "sct_singleCell.h5ad" and both are located in the proper folder. Is there something I'm doing wrong here? Thanks again!

[ankitbioinfo]

Hi, Thanks for your message. Can you also quickly check how many genes are overlapping between scRNAseq and spatial data?

[reinertanalytics]

Hi, thanks for the quick response! I only have 300 overlapping genes.

Number of overlapping genes: 300
Number of genes unique to spatial data: 0
Number of genes unique to RNA-seq data: 36288

[ankitbioinfo]

Thanks for quick answer. Actually the gene_covariation_analysis uses the raw counts matrix from "anndata.raw" layer.
The anndata.X that is normalized count only used in annotation step not in gene covariation step. Please check "sct_spatial.h5ad" and "sct_singleCell.h5ad" both should have raw counts in their raw layer. You can also check "Start_Data_prep.ipynb" line no 13 following these commands.

ad_spatial_common.raw=temp_spatial.copy()
ad_seq_common.raw=temp_seq.copy()

I hope it should work.