[results] different input files or loom versions having different auccell loom results

[lexuziandyu]

My version is:

pyscenic==0.12.1 loompy==2.0.17 anndata==0.9.2 scanpy==1.9.8

Compared with my previous pyscenic output file, the loom lacks the "MetaData" info.

Previous:

Class: H5Group
Filename: count.loom
Group: /attrs
Listing:
name obj_type dataset.dims dataset.type_class
CreationDate H5I_DATASET 0 H5T_STRING
LOOM_SPEC_VERSION H5I_DATASET 0 H5T_STRING
MetaData H5I_DATASET 0 H5T_STRING

Now:

Class: H5Group
Filename: celltype1.loom
Group: /attrs
Listing:
name obj_type dataset.dims dataset.type_class
LOOM_SPEC_VERSION H5I_DATASET 1 H5T_STRING

These two process are also different in: the previous one using a large SeuratObj, converted to a loom file, containing all celltypes, while the new one is subseted and converted to loom file only 1 celltype within.
The new output is a little bit different. When I perform

regulonThresholds = get_regulon_thresholds(loom)

It shows that:

Error in [[.H5Group(loom[["attrs"]], "MetaData") :
An object with name MetaData does not exist in this group

However, I can perform get_regulons_AUC and other functions on the Large aucellRseults properly. Is this possible that the input only-1-celltype loom file is too small for aucell step to compute the thresholds, or I need change my version to get the thresholds?

[lexuziandyu]

Update: after re-performing on the all-celltype-data, there is still a lack of "MetaData" slot. So it is believed that the problem is about the version.