www.guidetopharmacology.org Will the real targets please stand up ? Chris Southan PHAR/BPS Guide to PHARMACOLOGY Web portal Group, Centre for Integrative Physiology, School of Biomedical Sciences, University of Edinburgh, Hugh Robson Building, Edinburgh, EH8 9XD, UK. [email protected]1
Presented at ICCS, June 1-5 2014, Noordwijkerhout, The Netherlands, http://www.int-conf-chem-structures.org/
Will the real drug targets please stand up?
Discerning the molecular mechanisms of action (mmoa) for drugs treating human diseases is crucially important. This presentation will provide an overview of target numbers in IUPHAR/BPS Guide to PHARMACOLOGY, the curatorial challenges and compare these to other sources and consider the wider implications for drug discovery. We have developed stringent mapping criteria for primary targets (i.e. identifying those direct protein interactions mechanistically causative for therapeutic efficacy). This includes inter-source corroboration by intersecting multiple drug sources inside PubChem to produce consensus structure sets. An analogous approach is used to intersect published target lists and database subsets at the UniProtKB/Swiss-Prot identity level (a selection of drug and target lists is now hosted on our website http://www.guidetopharmacology.org/lists.jsp) . Our cumulative curation results reveal that structure representation differences, data provenance and variability of assay results, are major issues for experimental pharmacology and global database quality. While our activity mappings encompass some polypharmacolgy (e.g. dual inhibitors and kinase panel screens) our strategic choice is to annotate minimal, rather than maximal target sets. The consequent increased precision gives our database high utility for data mining, linking and cross-referencing. Our own database figures are currently converging to ~200 human protein primary targets for ~900 consensus chemical structures of approved small-molecule drugs. Target lists from other sources are typically larger. Comparative analysis of these lists by their UniProt ID content and Gene Ontology distributions suggests curatorial differences are the main cause of divergence . The global target landscape thus shows paradoxical trends. On the one hand, cumulative drug research output and recent expansions (e.g. epigenetic targets and orphan diseases) have pushed bioactive compounds from papers or patents to above 2 million and chemically modulatable human proteins above 1500. On the other hand, reports of Phase II clinical efficacy failure, with implicit target de-validation, are frequent. In addition, our assessment of drug approval data from 2009 to 2013 indicates new targets (i.e. first-in-class mmoas) are so low as to threaten the sustainability of the pharmaceutical industry. Causes and consequences of these paradoxes, along with utilities for minimal and maximal druggable genomes, will be discussed.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
www.guidetopharmacology.org
Will the real targets please stand up ?
Chris Southan
IUPHAR/BPS Guide to PHARMACOLOGY Web portal Group, Centre for Integrative Physiology, School of Biomedical Sciences, University of Edinburgh,
• Focus on minimal, rather than maximal relationship capture, to produce a more concise “drugged genome”
• Stringent primary activity mapping by citable results (e.g. Kd, Ki, IC50)
• Read the papers to resolve the results• Mask nutraceuticals/metabolites from drug interaction space• Use consensus target (UniProt IDs) as curation starting points• Use consensus drug structures (PubChem CIDs) as curation
starting points• Minimise complex subunit mapping to direct interactions• Don’t use matrix screen results for primary mappings• Human targets only (currently), mostly small molecules plus Abs• Pragmatic flexibility i.e. can include multi-mapping, dual
inhibitors, proven polypharmacology and unknown mmoas
Atorvastatin: mapped to different targets in 4 dbs
14
Atorvastatin vs DPPIV
15
Current GToPdb content
281 primary targets of approved drugs501 protein mappings of approved drugs
354 UniProt intersect
16
Conclusions
• There are many reasons why drug target lists are discordant
• It is thus useful to have many to compare and discern a consensus (i.e. getting the real ones to stand up)
• At GToPdb we use consensuses as starting points to activity-map a minimal set of targets
• Utility of maximal sets include possible polypharmacology and genetic associations
• Utility of minimal sets include defining basic mmoas, a core drugged genome, a pocketome , defining data gaps, and as “small (but perfectly formed) data” to underpin “big (noisy) data”
• First-in class expansions of the minimal set are perilously low
17
Acknowledgments and
references
Post-conference note: Organisations wishing to integrate GToPdb records are welcome to contact us