データ統合と計算の融合による創薬研究 水口賢司.

Transcription

1 データ統合と計算の融合による創薬研究 水口賢司

2 Outline - Phenotypic and target-based screening with ligand-based and structure-based drug design - Systems biology based approach - Data integration - Application to cancer drug discovery

3 Phenotypic screening and targetbased screening

4 Ligand based drug design (LBDD) Actives A large compound library Fingerprints Phenotypic screening Pharmacophore Inactives QSAR

5 Structure based drug design (SBDD) Target A large compound library/database Docking Scoring List of compounds for assays

6 How SBDD can assist target-based screening e.g., FXa inhibitors P1 N N N A basic fragment at P1 site was shown to be essential for mimicking the natural substrate by ligand based approaches Dabigatran Ligand based N O N N O N Structure based Rivaroxaban O Unfavorable ADMET profiles (Poor oral bioavailability) A neutral fragment Favorable ADMET profiles SBDD solved the problem with LBDD

7 Fundamental limitations of structurebased approaches SBDD Hit Multi-targeting activities Adverse event Target

8 The systems biology approach to drug discovery Source: Ingenuity Pathway Analysis, Causal Network Analysis suite Prathipati and Mizuguchi, Current Topics in Medicinal Chemistry (in press)

9 Data integration Pathway Protein-compound interactions (Predicted) structure Gene Protein Chemical compound Integration using InChI PubChem PubChem PubChem 3016 PubChem AAOVKJBEBIDNHE Diazepam C 16 H 13 ClN 2 O DrugBank DB00829 ChEBI PDB DZP DrugBank DB07699

10 Data warehouse system A Data warehouse compiles the contents of multiple databases to fit a common data model (Stein 2003, Nat Rev Genetics 4, )

11 Pathway Toxygates: an integrated platform for toxicogenomics data analysis RDF (remote/local) Gene, protein, pathway,... Gene Gene Probe Dynamic lookup from remote servers Probe Nystrom et al., Bioinformatics (2013) Compound A Time Dose Time Dose Probes ( 30,000) Sample Compound B Sample Samples ( 24,000) ( ranking ) Data matrix (key-value store, Kyoto Cabinet) (microarray data)

12 TargetMine Data warehouse for drug target prioritization Integrate a wide range of data sources. Enable complicated searches that are difficult to achieve with existing tools. Custom annotations and in-house data. Powered by at the Cambridge Systems Biology Centre Chen et al., PLoS ONE 6(3): e17844 (2011)

13 Data sources in TargetMine micro RNA & target associa7ons Transcrip7on factor target associa7ons Pathways & interac7ons Genes & proteins irefindex Domains & 3D structures Chemical compounds & interac7ons INTEGRATION Homo sapiens, Rattus norvegicus, Mus musculus

14 How InterMine integrates data Drug Protein Protein Gene Disease Ontology Gene Ontology Protein Gene Gene Pathway Drug Gene Ontology Protein Post-processing Gene Disease Ontology Pathway Gene Ontology (Java, PostgreSQL)

15 Given a protein, find all the compounds that target this protein Query protein: Catechol O-methyltransferase (COMT_HUMAN) 14 interac7ng compounds 36 non- redundant interac7ng compounds from ChEMBL, DrugBank and PDB 7 interac7ng compounds 17 interac7ng compounds ChEMBL DrugBank 2 5 PDB

16 Given a drug, find all the ChEMBL targets that have been co-crystalized with that drug Found 4 proteins

17 Targeting cancer stem cells using a systems biology approach Philip Prathipati Chiba et al., Scientific Reports (in press)