Skip to content

Commit 759d3d3

Browse files
npalacioescatnpalacioescat
committed
Replaced URL for data source in Hormone2Cell
1 parent 55178a5 commit 759d3d3

1 file changed

Lines changed: 30 additions & 7 deletions

File tree

‎pypath/inputs_v2/hormone2cell.py‎

Lines changed: 30 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -36,7 +36,15 @@
3636

3737
# =================================== SET-UP ===================================
3838

39-
URL = 'https://hormonecellatlas.org.uk/accounts/downloads/file/table-s2b/'
39+
URL = 'https://www.science.org/doi/suppl/10.1126/science.aeb2672/suppl_file/science.aeb2672_tables_s1_to_s11.zip'
40+
41+
files_skiprows = {
42+
'table-s2b': 3,
43+
'table-s2c': 2,
44+
'table-s2d': 2,
45+
'table-s2e': 3,
46+
'table-s2f': 2,
47+
}
4048

4149
config = ResourceConfig(
4250
id=ResourceCv.HORMONE2CELL,
@@ -53,19 +61,19 @@
5361
),
5462
)
5563

56-
5764
download = Download(
5865
url=URL,
59-
filename='hormone2cell.xlsx',
66+
filename='science.aeb2672_tables_s1_to_s11.zip',
6067
subfolder='hormone2cell',
6168
large=True,
62-
ext='xlsx',
69+
ext='zip',
6370
default_mode='rb',
6471
)
72+
# Fix download, cell/tissue info in table 3 and on?
6573

66-
def parser(opener):
74+
def parser(opener, skiprows):
6775

68-
df = pd.read_excel(opener.path, skiprows=3)
76+
df = pd.read_excel(opener.path, skiprows=skiprows)
6977

7078
yield from df.to_dict(orient='records')
7179

@@ -77,7 +85,7 @@ def parser(opener):
7785
transform={},
7886
)
7987

80-
schema = EntityBuilder(
88+
schema_2B = EntityBuilder(
8189
#entity_type=EntityTypeCv.
8290
)
8391

@@ -94,6 +102,7 @@ def parser(opener):
94102

95103
# ================================= REFERENCE ==================================
96104
# S1B
105+
#
97106
# Tissue System Classical endocrine cell type Neuroendocrine cell type Lineage Tissue level broad-grained cell type (celltype_level1) Tissue level fine-grained cell type (celltype_level2) Tissue level Cell Type Broad 1 Tissue level Cell Type Broad 2 Tissue level Cell Type Broad 3 Celltype_count
98107
# Fallopian Tube Reproductive not not epithelial epithelial_cell_nonciliated epithelial_cell_nonciliated_1 epithelial epithelial_cell_nonciliated epithelial_cell_nonciliated 7477
99108
# Fallopian Tube Reproductive not not immune t_cell t_cell_cd4 T_cell T_cell CD4_T_cell 5782
@@ -102,7 +111,18 @@ def parser(opener):
102111
# Fallopian Tube Reproductive not not mesenchymal myofibroblast myofibroblast_3 myofibroblast myofibroblast myofibroblast 4863
103112
# Fallopian Tube Reproductive not not epithelial epithelial_cell_nonciliated epithelial_cell_nonciliated_2 epithelial epithelial_cell_nonciliated epithelial_cell_nonciliated 7786
104113

114+
# S2B
115+
# X Y
116+
# hormone_ID hormone_ID_unique hormone_short hormone_figures hormone_display hormone_other Tier include exclude classical regulatory gut_pancreatic hormones neuropeptides prostaglandins reprodutive cardiovascular other ref_hormone ref_receptor
117+
# entry_001 hormone_001 acth ACTH ACTH Adrenocorticotropic hormone 1 1 NA 1 1 NA 1 NA NA NA NA https://pubmed.ncbi.nlm.nih.gov/30156493/ https://pubmed.ncbi.nlm.nih.gov/26793988/ https://pubmed.ncbi.nlm.nih.gov/28220105/
118+
# entry_002 hormone_002 activin_a Activin A Activin A   2 1 NA NA NA NA NA NA 1 NA NA https://pubmed.ncbi.nlm.nih.gov/27328872/ https://pubmed.ncbi.nlm.nih.gov/3012369/ https://pubmed.ncbi.nlm.nih.gov/30540228/ https://pubmed.ncbi.nlm.nih.gov/1646080/ https://pubmed.ncbi.nlm.nih.gov/15196700/ https://pubmed.ncbi.nlm.nih.gov/19273500/ https://pubmed.ncbi.nlm.nih.gov/26884470/
119+
# entry_003 hormone_003 activin_ab Activin AB Activin AB   2 1 NA NA NA NA NA NA NA NA 1 https://pubmed.ncbi.nlm.nih.gov/3086749/ https://pubmed.ncbi.nlm.nih.gov/27328872/ https://pubmed.ncbi.nlm.nih.gov/15196700/ https://pubmed.ncbi.nlm.nih.gov/26884470/
120+
# entry_004 hormone_004 activin_b Activin B Activin B   2 1 NA NA NA NA NA NA NA NA 1 https://pubmed.ncbi.nlm.nih.gov/1644823/ https://pubmed.ncbi.nlm.nih.gov/27328872/ https://pubmed.ncbi.nlm.nih.gov/15196700/ https://pubmed.ncbi.nlm.nih.gov/26884470/
121+
# entry_005 hormone_005 activin_e Activin E Activin E   3 1 NA NA NA NA NA NA NA NA NA https://pubmed.ncbi.nlm.nih.gov/8941337/ https://pubmed.ncbi.nlm.nih.gov/38533769/ https://pubmed.ncbi.nlm.nih.gov/38533769/
122+
123+
105124
# S2C
125+
# Z X*
106126
# gene category ID status_in_h2c
107127
# POMC hormone hormone_1 h2c_included
108128
# INHBA hormone hormone_2 h2c_included
@@ -111,6 +131,7 @@ def parser(opener):
111131
# C1QTNF12 hormone hormone_5 h2c_included
112132

113133
# S2D
134+
# Y
114135
# hormone_short hpc_include1 hpc_include2 hpc_include3 hpc_include4 hpc_include5 hpc_include6 hpc_include7 hpc_include8 hpc_exclude1 hpc_exclude2 receptor_known hormone_type_broad hormone_type_fine
115136
# adm2 ADM2 yes peptide_protein peptide
116137
# adrenaline TH DBH PNMT SLC18A1 yes amine_derived amine_derived
@@ -124,6 +145,7 @@ def parser(opener):
124145
# androgens STAR CYP11A1 FDX1 FDXR CYP17A1 CYB5A CYP19A1 CYP21A2 yes steroid steroid
125146

126147
# S2E
148+
# Y
127149
# hormone_short receptorgene1 receptorgene2 gene1_name gene2_name receptor_type_broad receptor_status
128150
# acth MC2R melanocortin 2 receptor gpcr established
129151
# activin_a ACVR2A ACVR1B Activin A receptor_type2A Activin A receptor_type1B enzyme_linked established
@@ -140,6 +162,7 @@ def parser(opener):
140162
# activin_e ACVR2B ACVR1C Activin A receptor_type2B Activin A receptor_type1C enzyme_linked established
141163

142164
# S2F
165+
# Z
143166
# gene_id gene_name uniprot_id h2c_present type analysis_cat reaction RHEA other genes other_name
144167
# CYP11A1 Cholesterol side-chain cleavage enzyme, mitochondrial P05108 1 enzyme_synthesis synthesis Pregnenolone_from_cholesterol Rhea:35739 FDX1, FDXR P450scc
145168
# CYP11B1 Cytochrome P450 11B1, mitochondrial P15538 1 enzyme_synthesis synthesis Corticosterone_from_11deoxycorticosterone Rhea:46104

0 commit comments

Comments
 (0)