Validator outputs misleading errors when NaN values are in required Visium .obs columns

[Jchaffer787]

If a NaN value is present in adata.obs['array_col'], ['array_row'], and/or ['in_tissue'], validator error output:

ERROR: obs['array_col'] is required for obs['assay_ontology_term_id'] 'EFO:0010961' (Visium Spatial Gene Expression) and uns['spatial']['is_single'] is True.

ERROR: obs['array_row'] is required for obs['assay_ontology_term_id'] 'EFO:0010961' (Visium Spatial Gene Expression) and uns['spatial']['is_single'] is True.

ERROR: obs['in_tissue'] is required for obs['assay_ontology_term_id'] 'EFO:0010961' (Visium Spatial Gene Expression) and uns['spatial']['is_single'] is True.

Implies fields are absent, instead of pointing towards the value issue.

[ejmolinelli]

We are now differentiating between TISSUE_POSITION_COL missing altogether vs having null/NaN values.

We have added an additional error message for null/NaN values which reads (for TISSUE_POSITION_COL="array_col"):

obs['array_col'] cannot have missing or NaN values when obs['assay_ontology_term_id'] is a descendant of 'EFO:0010961' (Visium Spatial Gene Expression) and uns['spatial']['is_single'] is True.