Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Compact models for compact devices: Visualisation of SAR using mobile apps

1,011 views

Published on

Presented at American Chemical Society meeting, Boston, 2015. Describes how cheminformatics algorithms and visualisation interfaces have advanced on mobile apps to cover a diverse variety of functionality, increasingly calculated on the device itself rather than deferring to a web service. Culminates in a demo of the PolyPharma app prototype (see http://cheminf20.org/2015/08/06/the-polypharma-app-a-mash-up-of-ideas-and-technology)

Published in: Science
  • Be the first to comment

  • Be the first to like this

Compact models for compact devices: Visualisation of SAR using mobile apps

  1. 1. Compact models for compact devices: Visualisation of SAR using mobile apps Alex M. Clark, Ph.D. August 2014 © 2015 Molecular Materials Informatics, Inc. http://molmatinf.com
  2. 2. MOLECULAR MATERIALS INFORMATICS Modern User Experience 2 • Continuity from 1980s • Mostly open architecture • Deploy on anything • Very open • Tied to server • Small touchscreen • Ultimate mobility • Unitary functionality • Company store
  3. 3. MOLECULAR MATERIALS INFORMATICS Mobile Particularly 3 Large Data Remote API Small Data Device Local • presentation • visualisation • calculation • sharing • searching • registration • online sharing • heavy calculations Medium Data • custom content • hard to appify
  4. 4. MOLECULAR MATERIALS INFORMATICS The Remoteness Problem 4 • Very tempting to use servers for heavy lifting • Requests via Internet API • User identity • Security issues • Maintenance burden • Hostile user experience - no signal - foreign countries - planes - wilderness
  5. 5. MOLECULAR MATERIALS INFORMATICS The Local Solution • Structure ➙ Property calculation algorithms • Start with scalars: - some are easy - others less so 5
  6. 6. MOLECULAR MATERIALS INFORMATICS Substructure Filters • PAINS and AZ Filters: both require complex recursive substructure queries • Overlay visualisation 6
  7. 7. MOLECULAR MATERIALS INFORMATICS Molecule Perception • Hardcoded algorithms for stereochemistry and tautomer enumeration • Displayed by structure annotation 7
  8. 8. MOLECULAR MATERIALS INFORMATICS Prepare & Shrink • Prepare data using desktop software (and/or of scripts) • Build Bayesian (ECFP6) model: concise, portable • Transfer model to mobile app... 8
  9. 9. MOLECULAR MATERIALS INFORMATICS Predict & Visualise • Fast to apply • Numeric or molecular visualisation • Overlay uses ECFP6 fingerprints • With one molecule or many 9
  10. 10. MOLECULAR MATERIALS INFORMATICS More is Better • Extract target:activity collections from ChEMBL • Compute active/inactive boundaries • Automatically build Bayesian models... 10 3 60 9 12 frequency activity optimal boundary even distribution JCIM 55, 1246-1260 (2015) Target β-Hydroxysteroid dehydrogenase 1-Acylglycerol-3-phosphate Acetylcholine esterase ADAM10 Adenosine A1 receptor Adrenergic receptor α1 Aldoketo reductase Alkaline phosphatase Aminopeptidase N Angiotensin converting enzyme Anthrax lethal factor Bcl2 Arachidonate 12 ChEMBL ~2000
  11. 11. MOLECULAR MATERIALS INFORMATICS Extreme Appification • PolyPharma app: work in progress • Visualise a lot of SAR data... with 2 taps 11
  12. 12. MOLECULAR MATERIALS INFORMATICS Canned Demo 12
  13. 13. Acknowledgments http://molmatinf.com http://molsync.com http://cheminf20.org @aclarkxyz • Sean Ekins, Antony Williams, Andy Davis • Royal Society of Chemistry • Collaborative Drug Discovery • Inquiries to info@molmatinf.com

×