Merge pull request #51 from brain-bican/bug_fix

load from cas json

examples/human_m1/CCN202204130_project_config.yaml → examples/human_m1/CS202210140_project_config.yaml

@@ -1,9 +1,11 @@
-id: CCN202204130
+id: CS202210140
 title: Human M1 taxonomy
 description: Atlas of human primary motor cortex (M1), developed in collaboration with the BRAIN Initiative Cell Census Network (BICCN).
+matrix_file_id: CellXGene_dataset:Siletti_all_non_neuronal_cells_v2
 github_org: hkir-dev
 repo: human_m1
 author: https://orcid.org/0000-0002-7073-9172
+accession_id_prefix: CS202210140_
 citation: ''
 creators:
   - https://orcid.org/0000-0001-7258-9596

requirements.txt

@@ -11,6 +11,6 @@ openpyxl
 # git+https://github.com/ontodev/axle.git
 urlp
 mkdocs
-cas-tools==0.0.1.dev11
+cas-tools==0.0.1.dev14
 pandas
 tdta==0.1.0.dev2

scripts/import.py

@@ -8,6 +8,7 @@
 
 from cas.ingest.ingest_user_table import ingest_user_data
 from cas.flatten_data_to_tables import serialize_to_tables
+from cas.file_utils import read_cas_json_file
 from pathlib import Path
 from ruamel.yaml import YAML
 
@@ -48,17 +49,22 @@ def import_data(input, schema, curation_tables):
                 user_cas_path = f
                 new_files.append(user_cas_path)
 
-    if user_data_path:
-        user_data_ct_path = add_user_table_to_nanobot(user_data_path, schema, curation_tables)
-        new_files.append(user_data_ct_path)
+    # provide either json or tsv + yaml
+    if user_cas_path:
+        user_file_name = os.path.splitext(os.path.basename(user_cas_path))[0]
+        std_data = read_cas_json_file(user_cas_path)
     else:
-        raise Exception("Couldn't find the cell type annotation config file (with yaml or yml extension) in folder: " + input)
+        if user_data_path:
+            user_data_ct_path = add_user_table_to_nanobot(user_data_path, schema, curation_tables)
+            new_files.append(user_data_ct_path)
+        else:
+            raise Exception("Couldn't find the cell type annotation config file (with yaml or yml extension) in folder: " + input)
 
-    if user_config_path:
-        user_file_name = os.path.splitext(os.path.basename(user_data_path))[0]
-        std_data = ingest_user_data(user_data_path, user_config_path)
-    else:
-        raise Exception("Couldn't find the config data files (with yaml or yml extension) in folder: " + input)
+        if user_config_path:
+            user_file_name = os.path.splitext(os.path.basename(user_data_path))[0]
+            std_data = ingest_user_data(user_data_path, user_config_path)
+        else:
+            raise Exception("Couldn't find the config data files (with yaml or yml extension) in folder: " + input)
 
     accession_prefix = retrieve_accession_prefix(Path(input).parent.absolute())
     std_tables = serialize_to_tables(std_data, user_file_name, input, accession_prefix)