Cellxgene export

sfaira/consts/adata_fields.py

@@ -17,7 +17,8 @@ class AdataIds:
     cellontology_id: str
     development_stage: str
     disease: str
-    doi: str
+    doi_journal: str
+    doi_preprint: str
     download_url_data: str
     download_url_meta: str
     dataset: str
@@ -87,7 +88,8 @@ def __init__(self):
         self.cellontology_id = "cell_ontology_id"
         self.default_embedding = "default_embedding"
         self.disease = "disease"
-        self.doi = "doi"
+        self.doi_journal = "doi_journal"
+        self.doi_preprint = "doi_preprint"
         self.dataset = "dataset"
         self.dataset_group = "dataset_group"
         self.download_url_data = "download_url_data"
@@ -123,6 +125,7 @@ def __init__(self):
         self.unknown_celltype_identifier = "UNKNOWN"
         self.not_a_cell_celltype_identifier = "NOT_A_CELL"
         self.unknown_metadata_identifier = "unknown"
+        self.unknown_metadata_ontology_id_identifier = "unknown"
 
         self.obs_keys = [
             "assay_sc",
@@ -152,7 +155,8 @@ def __init__(self):
             "annotated",
             "author",
             "default_embedding",
-            "doi",
+            "doi_journal",
+            "doi_preprint",
             "download_url_data",
             "download_url_meta",
             "id",
@@ -182,9 +186,11 @@ def __init__(self):
         self.cellontology_class = "cell_type"
         self.cellontology_id = "cell_type_ontology_term_id"
         self.default_embedding = "default_embedding"
-        self.doi = "preprint_doi"
+        self.doi_journal = "publication_doi"
+        self.doi_preprint = "preprint_doi"
         self.disease = "disease"
         self.gene_id_symbols = "gene_symbol"
+        self.gene_id_ensembl = "ensembl"
         self.gene_id_index = self.gene_id_symbols
         self.id = "id"
         self.ncells = "ncells"
@@ -228,11 +234,13 @@ def __init__(self):
             "tech_sample",
         ]
         self.var_keys = [
+            "gene_id_ensembl",
             "gene_id_symbols",
         ]
         self.uns_keys = [
+            "doi_journal",
+            "doi_preprint",
             "default_embedding",
-            "id",
             "title",
         ]
         # These attributes related to obs and uns keys above are also in the data set attributes that can be

sfaira/consts/ontologies.py

@@ -24,7 +24,8 @@ def __init__(self):
             "mouse": OntologyMmusdv(),
         }
         self.disease = OntologyMondo()
-        self.doi = None
+        self.doi_journal = None
+        self.doi_preprint = None
         self.ethnicity = {
             "human": None,  # TODO OntologyHancestro
             "mouse": None,

sfaira/data/base/dataset_group.py

@@ -561,11 +561,11 @@ def doi(self) -> List[str]:
         """
         dois = []
         for _, v in self.datasets.items():
-            vdoi = v.doi
+            vdoi = v.doi_journal
             if isinstance(vdoi, str):
                 vdoi = [vdoi]
             dois.extend(vdoi)
-        return np.sort(np.unique(vdoi)).tolist()
+        return np.sort(np.unique(dois)).tolist()
 
     @property
     def supplier(self) -> List[str]:

sfaira/data/dataloaders/loaders/d10_1016_j_cell_2017_09_004/human_isletoflangerhans_2017_smartseq2_enge_001.py

@@ -18,7 +18,8 @@ def __init__(self, **kwargs):
 
         self.author = "Enge"
         self.disease = "healthy"
-        self.doi = "10.1016/j.cell.2017.09.004"
+        self.doi_journal = "10.1016/j.cell.2017.09.004"
+        self.doi_preprint = "10.1101/108043"
         self.normalization = "raw"
         self.assay_sc = "Smart-seq2"
         self.organ = "islet of Langerhans"

sfaira/data/dataloaders/loaders/d10_1016_j_cell_2018_02_001/mouse_x_2018_microwellseq_han_x.py

@@ -301,7 +301,7 @@ def __init__(self, **kwargs):
         self.author = "Han"
         self.dev_stage = sample_dev_stage_dict[self.sample_fn]
         self.disease = "healthy"
-        self.doi = "10.1016/j.cell.2018.02.001"
+        self.doi_journal = "10.1016/j.cell.2018.02.001"
         self.normalization = "raw"
         self.organism = "mouse"
         self.assay_sc = "microwell-seq"

sfaira/data/dataloaders/loaders/d10_1016_j_cell_2018_08_067/human_laminapropriaofmucosaofcolon_2019_10xsequencing_kinchen_001.yaml

@@ -7,9 +7,9 @@ dataset_wise:
     author:
         - "Kinchen"
     default_embedding:
-    doi:
-        - "10.1016/j.cell.2018.08.067"
-    download_url_data:
+    doi_journal: "10.1016/j.cell.2018.08.067"
+    doi_preprint:
+    download_url_data: 
         HC: "https://www.ncbi.nlm.nih.gov/geo/download/?acc=GSE114374&format=file&file=GSE114374%5FHuman%5FHC%5Fexpression%5Fmatrix%2Etxt%2Egz"
         UC: "https://www.ncbi.nlm.nih.gov/geo/download/?acc=GSE114374&format=file&file=GSE114374%5FHuman%5FUC%5Fexpression%5Fmatrix%2Etxt%2Egz"
     download_url_meta:

sfaira/data/dataloaders/loaders/d10_1016_j_cell_2019_06_029/human_colonicepithelium_2019_10xsequencing_smilie_001.py

@@ -16,7 +16,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "10x technology"
         self.author = "Smilie"
         self.disease = "healthy"
-        self.doi = "10.1016/j.cell.2019.06.029"
+        self.doi_journal = "10.1016/j.cell.2019.06.029"
         self.normalization = "raw"
         self.organ = "colonic epithelium"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1016_j_cell_2019_08_008/human_ileum_2019_10xsequencing_martin_001.py

@@ -16,7 +16,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "10x technology"
         self.author = "Martin"
         self.disease = "healthy"
-        self.doi = "10.1016/j.cell.2019.08.008"
+        self.doi_journal = "10.1016/j.cell.2019.08.008"
         self.normalization = "raw"
         self.organ = "ileum"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1016_j_celrep_2018_11_086/human_prostategland_2018_10xsequencing_henry_001.py

@@ -21,7 +21,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "10x technology"
         self.author = "Henry"
         self.disease = "healthy"
-        self.doi = "10.1016/j.celrep.2018.11.086"
+        self.doi_journal = "10.1016/j.celrep.2018.11.086"
         self.normalization = "raw"
         self.sample_source = "primary_tissue"
         self.state_exact = "healthy"

sfaira/data/dataloaders/loaders/d10_1016_j_cels_2016_08_011/human_pancreas_2016_indrop_baron_001.py

@@ -19,7 +19,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "inDrop"
         self.author = "Baron"
         self.disease = "healthy"
-        self.doi = "10.1016/j.cels.2016.08.011"
+        self.doi_journal = "10.1016/j.cels.2016.08.011"
         self.normalization = "raw"
         self.organ = "pancreas"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1016_j_cmet_2016_08_020/human_pancreas_2016_smartseq2_segerstolpe_001.py

@@ -17,7 +17,7 @@ def __init__(self, **kwargs):
         self.download_url_meta = "https://www.ebi.ac.uk/arrayexpress/files/E-MTAB-5061/E-MTAB-5061.sdrf.txt"
 
         self.author = "Segerstolpe"
-        self.doi = "10.1016/j.cmet.2016.08.020"
+        self.doi_journal = "10.1016/j.cmet.2016.08.020"
         self.normalization = "raw"
         self.organ = "pancreas"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1016_j_cmet_2019_01_021/mouse_pancreas_2019_10xsequencing_thompson_x.py

@@ -29,7 +29,7 @@ def __init__(self, **kwargs):
         self.download_url_meta = f"private,{self.sample_fn}_annotation.csv"
 
         self.author = "Thompson"
-        self.doi = "10.1016/j.cmet.2019.01.021"
+        self.doi_journal = "10.1016/j.cmet.2019.01.021"
         self.normalization = "raw"
         self.organ = "pancreas"
         self.organism = "mouse"

sfaira/data/dataloaders/loaders/d10_1016_j_devcel_2020_01_033/human_lung_2020_10xsequencing_miller_001.py

@@ -16,7 +16,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "10x technology"
         self.author = "Miller"
         self.disease = "healthy"
-        self.doi = "10.1016/j.devcel.2020.01.033"
+        self.doi_journal = "10.1016/j.devcel.2020.01.033"
         self.normalization = "raw"
         self.organ = "lung"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1016_j_neuron_2019_06_011/human_brain_2019_dropseq_polioudakis_001.yaml

@@ -4,9 +4,9 @@ dataset_structure:
 dataset_wise:
     author:
         - "Polioudakis"
-    doi:
-        - "10.1016/j.neuron.2019.06.011"
-    download_url_data:
+    doi_journal: "10.1016/j.neuron.2019.06.011"
+    doi_preprint:
+    download_url_data: 
         - "manual,sc_dev_cortex_geschwind.zip,http://solo.bmap.ucla.edu/shiny/webapp"
     download_url_meta:
     normalization: "raw"

sfaira/data/dataloaders/loaders/d10_1038_nmeth_4407/human_brain_2017_droncseq_habib_001.py

@@ -16,7 +16,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "DroNc-seq"
         self.author = "Habib"
         self.disease = "healthy"
-        self.doi = "10.1038/nmeth.4407"
+        self.doi_journal = "10.1038/nmeth.4407"
         self.normalization = "raw"
         self.organ = "brain"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1038_s41422_018_0099_2/human_testis_2018_10xsequencing_guo_001.py

@@ -16,7 +16,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "10x technology"
         self.author = "Guo"
         self.disease = "healthy"
-        self.doi = "10.1038/s41422-018-0099-2"
+        self.doi_journal = "10.1038/s41422-018-0099-2"
         self.normalization = "raw"
         self.organ = "testis"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1038_s41467_018_06318_7/human_caudatelobeofliver_2018_10xsequencing_macparland_001.py

@@ -15,7 +15,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "10x technology"
         self.author = "MacParland"
         self.disease = "healthy"
-        self.doi = "10.1038/s41467-018-06318-7"
+        self.doi_journal = "10.1038/s41467-018-06318-7"
         self.normalization = "raw"
         self.organ = "caudate lobe of liver"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1038_s41467_019_10861_2/human_kidney_2019_droncseq_lake_001.py

@@ -17,7 +17,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "DroNc-seq"
         self.author = "Lake"
         self.disease = "healthy"
-        self.doi = "10.1038/s41467-019-10861-2"
+        self.doi_journal = "10.1038/s41467-019-10861-2"
         self.normalization = "raw"
         self.organ = "kidney"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1038_s41467_019_12464_3/human_x_2019_10xsequencing_szabo_001.py

@@ -60,7 +60,7 @@ def __init__(self, **kwargs):
 
         self.assay_sc = "10x technology"
         self.author = "Szabo"
-        self.doi = "10.1038/s41467-019-12464-3"
+        self.doi_journal = "10.1038/s41467-019-12464-3"
         self.individual = SAMPLE_DICT[self.sample_fn][1]
         self.normalization = "raw"
         self.organ = SAMPLE_DICT[self.sample_fn][0]

sfaira/data/dataloaders/loaders/d10_1038_s41467_019_12780_8/human_retina_2019_10xsequencing_menon_001.py

@@ -14,7 +14,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "10x technology"
         self.author = "Menon"
         self.disease = "healthy"
-        self.doi = "10.1038/s41467-019-12780-8"
+        self.doi_journal = "10.1038/s41467-019-12780-8"
         self.normalization = "raw"
         self.organ = "retina"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1038_s41586_018_0698_6/human_placenta_2018_x_ventotormo_001.py

@@ -22,7 +22,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "10x technology" if self.sample_fn == "E-MTAB-6678.processed" else "Smart-seq2"
         self.author = "Ventotormo"
         self.disease = "healthy"
-        self.doi = "10.1038/s41586-018-0698-6"
+        self.doi_journal = "10.1038/s41586-018-0698-6"
         self.normalization = "raw"
         self.organ = "placenta"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1038_s41586_019_1373_2/human_liver_2019_celseq2_aizarani_001.py

@@ -15,7 +15,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "CEL-seq2"
         self.author = "Aizarani"
         self.disease = "healthy"
-        self.doi = "10.1038/s41586-019-1373-2"
+        self.doi_journal = "10.1038/s41586-019-1373-2"
         self.normalization = "raw"
         self.sample_source = "primary_tissue"
         self.organ = "liver"

sfaira/data/dataloaders/loaders/d10_1038_s41586_019_1631_3/human_liver_2019_10xsequencing_ramachandran_001.py

@@ -16,7 +16,7 @@ def __init__(self, **kwargs):
 
         self.assay_sc = "10x technology"
         self.author = "Ramachandran"
-        self.doi = "10.1038/s41586-019-1631-3"
+        self.doi_journal = "10.1038/s41586-019-1631-3"
         self.normalization = "raw"
         self.organ = "liver"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1038_s41586_019_1652_y/human_liver_2019_10xsequencing_popescu_001.py

@@ -14,7 +14,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "10x technology"
         self.author = "Popescu"
         self.disease = "healthy"
-        self.doi = "10.1038/s41586-019-1652-y"
+        self.doi_journal = "10.1038/s41586-019-1652-y"
         self.normalization = "raw"
         self.organ = "liver"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1038_s41586_019_1654_9/human_brain_2019_10x3v2sequencing_kanton_001.yaml

@@ -4,9 +4,9 @@ dataset_structure:
 dataset_wise:
     author:
         - "Kanton"
-    doi:
-        - "10.1038/s41586-019-1654-9"
-    download_url_data:
+    doi_journal: "10.1038/s41586-019-1654-9"
+    doi_preprint:
+    download_url_data: 
         - "https://www.ebi.ac.uk/arrayexpress/files/E-MTAB-7552/E-MTAB-7552.processed.3.zip"
     download_url_meta:
         - "https://www.ebi.ac.uk/arrayexpress/files/E-MTAB-7552/E-MTAB-7552.processed.1.zip"

sfaira/data/dataloaders/loaders/d10_1038_s41586_020_2157_4/human_x_2020_microwellseq_han_x.py

@@ -19,7 +19,7 @@ def __init__(self, **kwargs):
         ]
 
         self.author = "Han"
-        self.doi = "10.1038/s41586-020-2157-4"
+        self.doi_journal = "10.1038/s41586-020-2157-4"
         self.healthy = True
         self.normalization = "raw"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1038_s41586_020_2922_4/human_lung_2020_x_travaglini_001.yaml

@@ -7,9 +7,9 @@ dataset_wise:
     author:
         - "Travaglini"
     default_embedding: "X_tSNE"
-    doi:
-        - "10.1038/s41586-020-2922-4"
-    download_url_data:
+    doi_journal: "10.1038/s41586-020-2922-4"
+    doi_preprint:
+    download_url_data: 
         droplet_normal_lung_blood_scanpy.20200205.RC4.h5ad: "syn21625095,droplet_normal_lung_blood_scanpy.20200205.RC4.h5ad"
         facs_normal_lung_blood_scanpy.20200205.RC4.h5ad: "syn21625142,facs_normal_lung_blood_scanpy.20200205.RC4.h5ad"
     download_url_meta:

sfaira/data/dataloaders/loaders/d10_1038_s41590_020_0602_z/human_colon_2020_10xsequencing_james_001.py

@@ -16,7 +16,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "10x technology"
         self.author = "James"
         self.disease = "healthy"
-        self.doi = "10.1038/s41590-020-0602-z"
+        self.doi_journal = "10.1038/s41590-020-0602-z"
         self.normalization = "raw"
         self.organ = "colon"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1038_s41591_019_0468_5/human_lung_2019_dropseq_braga_001.py

@@ -15,7 +15,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "Drop-seq"
         self.author = "Braga"
         self.disease = "healthy"
-        self.doi = "10.1038/s41591-019-0468-5"
+        self.doi_journal = "10.1038/s41591-019-0468-5"
         self.normalization = "raw"
         self.organ = "lung"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1038_s41591_019_0468_5/human_x_2019_10xsequencing_braga_x.py

@@ -20,7 +20,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "10x technology"
         self.author = "Braga"
         self.disease = "healthy"
-        self.doi = "10.1038/s41591-019-0468-5"
+        self.doi_journal = "10.1038/s41591-019-0468-5"
         self.normalization = "scaled"
         self.organ = "bronchus" if self.sample_fn == "vieira19_Bronchi_anonymised.processed.h5ad" else "lung parenchyma"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1038_s41593_019_0393_4/mouse_x_2019_10xsequencing_hove_001.py

@@ -19,7 +19,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "10x technology"
         self.author = "Hove"
         self.disease = "healthy"
-        self.doi = "10.1038/s41593-019-0393-4"
+        self.doi_journal = "10.1038/s41593-019-0393-4"
         self.normalization = "raw"
         self.organism = "mouse"
         self.sample_source = "primary_tissue"

sfaira/data/dataloaders/loaders/d10_1038_s41597_019_0351_8/human_kidney_2020_10xsequencing_liao_001.py

@@ -23,7 +23,7 @@ def __init__(self, **kwargs):
         self.organism = "human"
         self.sample_source = "primary_tissue"
         self.year = 2020
-        self.doi = "10.1038/s41597-019-0351-8"
+        self.doi_journal = "10.1038/s41597-019-0351-8"
 
         self.gene_id_symbols_var_key = "names"
         self.gene_id_ensembl_var_key = "ensembl"

sfaira/data/dataloaders/loaders/d10_1073_pnas_1914143116/human_retina_2019_10xsequencing_voigt_001.py

@@ -15,7 +15,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "10x technology"
         self.author = "Voigt"
         self.disease = "healthy"
-        self.doi = "10.1073/pnas.1914143116"
+        self.doi_journal = "10.1073/pnas.1914143116"
         self.normalization = "norm"
         self.organ = "retina"
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1084_jem_20191130/human_x_2019_10xsequencing_wang_001.py

@@ -24,7 +24,7 @@ def __init__(self, **kwargs):
         self.assay_sc = "10x technology"
         self.author = "Wang"
         self.disease = "healthy"
-        self.doi = "10.1084/jem.20191130"
+        self.doi_journal = "10.1084/jem.20191130"
         self.normalization = "raw"
         self.organ = organ
         self.organism = "human"

sfaira/data/dataloaders/loaders/d10_1101_2020_03_13_991455/human_lung_2020_10xsequencing_lukassen_001.py

@@ -21,7 +21,8 @@ def __init__(self, **kwargs):
         self.assay_sc = "10x technology"
         self.author = "Lukassen"
         self.disease = "healthy"
-        self.doi = "10.1101/2020.03.13.991455"
+        self.doi_journal = "10.15252/embj.20105114"
+        self.doi_preprint = "10.1101/2020.03.13.991455"
         self.normalization = "raw"
         self.organ = "lung"
         self.organism = "human"