IU Integrative Data Science Lab (IDSL) Type-2 Diabetes project
The output dataset produced by the IU Integrative Data Science Lab T2D project. In this CTSI-funded project, "Semantically mapping the hidden molecular connections of type-2 diabetes", Chem2Bio2RDF and SLAP were applied for knowledge discovery. We share the resulting dataset as a knowledge network (KN), for further use by the research community.
| type | count | notes |
|---|---|---|
| query compounds | 16 | T2D drugs, with PubChem CIDs |
| query targets | 2228 | C2B2R protein targets with SwissProt IDs |
| compounds | 289 | PubChem CIDs |
| associated targets | 71 | Associated by SLAP DT queries |
| GO | 39 | Gene Ontology class |
| chebi | 18 | |
| pathway | 12 | KEGG pathway |
| disease | 2 | OMIM disease |
| side effect | 3 | SIDER entity |
| substructure | 7 | chemical substructure |
| tissue | 1 | ~ |
| file | contents |
|---|---|
diabetes_drugs.smi |
16 SMILES, PubChem CIDs, and names |
protein_list.csv |
2228 C2B2R protein targets with SwissProt IDs |
t2d_dtp_links.csv |
SLAP summary results, 56 associations |
t2d_dtp.csv |
SLAP results, includes 56 hits and 35592 misses |
slap_dtp_merged.xgmml |
Merged, annotated, association network |
CID_TID.graphml |
(56 files) association networks for all SLAP hits |
| term | definition |
|---|---|
| T2D or T2DM | Type-2 Diabetes, a.k.a. Type-2 Diabetes Mellitus |
| C2B2R | Chem2Bio2RDF |
| SLAP | Semantically Linked Association Prediction |
| DTP | drug-target prediction (SLAP method) |
| KN or KG | knowledge network or graph |
| AN | association network, refers to SLAP sub-network associating one compound to one target, including linking associations |
From query compounds which are known diabetes drugs,heterogeneous linked datasets lead to novel targets via domain appropriate path patterns.
slap_dtp_merged_nodes_export.csv "class" counts:
"class" is column 5.
values, col 5 "class", unique: 10 ; total: 443
class counts
"GO" 39 ( 8.804%)
"chebi" 18 ( 4.063%)
"gene" 71 (16.027%)
"gene_family" 1 ( 0.226%)
"kegg_pathway" 12 ( 2.709%)
"omim_disease" 2 ( 0.451%)
"pubchem_compound" 289 (65.237%)
"sider" 3 ( 0.677%)
"substructure" 7 ( 1.580%)
"tissue" 1 ( 0.226%)
slap_dtp_merged_edges_export.csv "label" counts:
"label" is column 5.
values, col 5 "label", unique: 11 ; total: 4577
label counts
"GO_ID" 290 ( 6.336%)
"Gene_Family_Name" 32 ( 0.699%)
"chebi" 163 ( 3.561%)
"chemogenomics" 3630 (79.310%)
"cid" 52 ( 1.136%)
"drug" 31 ( 0.677%)
"expression" 146 ( 3.190%)
"hprd" 110 ( 2.403%)
"protein" 73 ( 1.595%)
"substructure" 48 ( 1.049%)
"tissue" 2 ( 0.044%)
Neo4j
has emerged as a leading graph database system,
combining concepts from Semantic Web Technologies,
including RDF Subject→Predicate→Object triples, with
the reliability and performance of enterprise software.
Although a commercial product, Neo4j
provides a free Community Edition, and supports open standards
such as Sparql queries and JSON results.
We have followed the example of other biomedical informatics
projects by converting the T2D-Net dataset
to Neo4j and sharing the workflow.
neo4j-client -i load_cytoscape_cyjs2tsv.cql -u neo4j -p ****** localhost
neo4j-client -i load_extras.cql -u neo4j -p ****** localhost
neo4j-client -i db_describe.cql -u neo4j -p ****** localhost