Contact ACS Webinars at Join a global community of over 150,000 chemistry professionals

Transcription

1 Have Questions? Type them into questions box! Why am I muted? Don t worry. Everyone is muted except the presenter and host. Thank you and enjoy the show. Contact ACS Webinars at acswebinars@acs.org 1 Join a global community of over 150,000 chemistry professionals Find the many benefits of ACS membership! 2 1

2 Benefits of ACS Membership Chemical & Engineering News (C&EN) The preeminent weekly digital and print news source. NEW! ACS SciFinder ACS Members receive 25 complimentary SciFinder research activities per year. NEW! ACS Career Navigator Your source for leadership development, professional education, career services, and much more. @AmerChemSociety Contact ACS Webinars at acswebinars@acs.org 4 2

3 All things around ACS Webinars are benefits! We learn about great research topics, we take advantage of our Membership, and we are being friendly to the environment because we can connect from anywhere. Jans Alzate-Morales Pharmaceutical Chemist, Associate Professor, and Director of Ph.D Program in Sciences Mention in Modeling of Chemical and Biological Systems, Centre for Bioinformatics and Molecular Simulations (CBSM), Universidad de Talca ACS member for 9 years strong! yvonnecmartin@comcast.net Be a featured fan on an upcoming webinar! Write to acswebinars@acs.org 5 Learn from the best and brightest minds in chemistry! Hundreds of webinars on diverse topics presented by experts in the chemical sciences and enterprise. Recordings are an exclusive ACS member benefit and are made available to registrants via an invitation once the recording has been edited and posted. Live Broadcasts of ACS Webinars continue to be available to the general public on Thursdays from 2-3pm ET! 6 3

4 What is ACS on Campus? ACS visits campuses across the world offering FREE seminars on how to be published, find a job, network and use essential tools like SciFinder. ACS on Campus presents seminars and workshops focused on how to: 7 #HeroesofChemistry ACS Heroes of Chemistry Award The ACS Heroes of Chemistry Award is the Annual award sponsored by the American Chemical Society that recognizes talented industrial chemical scientists whose work has led to the development of successful commercialized products ingrained with chemistry for the benefit of humankind Winners: 8 4

5 An individual development planning tool for you! 9 Upcoming ACS Webinar

6 11 Join the Division Today! For $25 ($10 for students), You Will Receive: A free digital copy of our annual medicinal chemistry review volume (over 680 pages, $160 retail price) Abstracts of MEDI programming at national meetings Access to student travel grants and fellowships Find out more about the ACS MEDI Division!

7 Available Now! Coming out in May 2019! 13 Celebrating 5 years & 50 Drug Discovery Webinars!

8 T H I S A C S W E B I N A R W I L L B E G I N S H O R T L Y... Effective Exploration of Chemical Space in Hit-Finding Johanna Hanneke Jansen Director, Novartis Institutes for BioMedical Research Zoe Cournia Investigator Assistant Professor Level, Biomedical Research Foundation, Academy of Athens and JCIM Associate Editor Slides available now! Recordings are an exclusive ACS member benefit. This ACS Webinar is co-produced with the ACS Division of Medicinal Chemistry, the American Association of Pharmaceutical Scientists, and the Journal of Chemical Information and Modeling 16 8

9 Chemical space is vast Image courtesy Peter Ertl, Novartis (2013) Hit-Finding: Assess a selection from chemical space in your assay Every well counts

10 Audience Survey Question ANSWER THE QUESTION ON BLUE SCREEN IN ONE MOMENT 19 Kingston Black is a cultivar of apple mostly used for? Apple pie Apple sauce Snacking Apple Cider Creative Commons photo by Andy Roberts * If your answer differs greatly from the choices above tell us in the chat! Quality means different things for different projects Highest quality apple? Apple pie Keep structure Apple sauce Cook up smooth Hiking snack Packable Apple cider Acidity and tannins 20 10

11 Quality means different things for different projects: hit-finding Cellular assay Non-cytotoxic Fragment-screen Solubility Fluorescent read-out No fluorophores MedChem starting point Novelty 21 Articulation of quality as it relates to purpose drives selection strategy Every well counts... Consider exclusions & bias Do not include compounds that are incompatible with quality & purpose Select highest quality compounds as representatives (highest likelihood of success) Availability of privileged core set? (computational model, related hit-list) Balance quality with diversity to cover chemical space Opportunity to discover novel mechanisms or chemical matter outside of current domain 22 11

12 Biased Complement Diversity, BCD Three internal case studies show that BCD works Bacteriology: Increased proportion of non-cytotoxic hits from 31% to 62% Two target-driven hit-finding campaigns identified quality starting points by complementing knowledge-driven subsets with biased diversity sets 23 Schematic representation of BCD Create compound library Available structures Chemical space: clustering of 2D fingerprints Quality attributes: Class A-D & exclusions Loop over every cluster & assess quality Keep all singletons If all-d: keep single best Else: 3-tier biased selection Biased selection User-defined coverage number RDKit Diversity Picker KNIME node Explicit complement: class-a is core-set (selected completely) 24 12

13 KNIME workflow with multiple meta-nodes for performing the selection 25 Time to run Malaria case study on Windows laptop (picking 50k from 300k molecules) 1 hour 20 minutes with predefined cluster-list (as deposited in SI) 3 hours when including clustering (ICM MolCluster node) Input and output nodes / notes Input 1: Structure input Available structures (after exclusions) Cluster information Classification: class A-D Input 2: Five variables Core-set? Sampling density Approach to pick single-best from all-d clusters Output: Identifiers of the selected subset Dedicate brainpower to decide and/or develop core-set(s) and quality attributes 26 13

14 Audience Survey Question ANSWER THE QUESTION ON BLUE SCREEN IN ONE MOMENT 27 Antibiotic drug discovery is more difficult for Gram-negative bacteria compared to Gram-positive bacteria mostly because of? Efflux The presence of a second, outer membrane Generic competition Resistance * If your answer differs greatly from the choices above tell us in the chat! Bacteriology case study: Optimize quality of hits that target Gram(-) pathogens Define quality: reduced cytotoxicity against mammalian cells Translate quality to attributes for classification: Analyze relationship between Gram(-) activity, clogd, and cytotoxicity (in-house data) 87% of compounds with activity < 20 M in Gram(-) and a clogd > 3 are cytotoxic vs mammalian cells Classification: Assign all compounds with clogd > 3 to class D Sliding scale bias for MW, ROT, PSA, HBD from class A C Additional bias using substructure flags Success criterion: Increase proportion of confirmed hits that are free of cytotoxicity Historical proportion is 31% for compounds with EC50 < 20 M in Gram(-) 28 14

15 Bacteriology BCD design increases the proportion of non-cytotoxic hits Confirmed hits: EC 50 < 20 M in 1 WT Gram(-) strain and no cytotoxicity against mammalian cells Historical proportion non-cytotoxic hits: 310/985 = 31% BCD proportion non-cytotoxic hits: 40/65 = 62% = success! Source set Screening set # Confirmed hits 832k 76k 40 Color by class A B C D 29 Perspective on using overall hit-rate to assess success: DON T! Comparable full deck data-set (bacteriology) Class # Compounds in screening set Class % in screening set # Primary hits Hit-rate A 205, , B 158, , C 81, , D 344, , Hit-rate is not constant across classes Total 789,620 42, BCD selection (bacteriology) Class # Compounds in screening set Class % in screening set # Primary hits Hit-rate A 50, B 10, C 1, D 8, BCD samples more from lower hit-rate class Total 70,904 1,

16 BCD increases the number of high quality hits Unbiased diversity* BCD Class Class % in set # Primary hits Hit-rate Class % in set # Primary hits Hit-rate A B C D 46 3, Overall 4, , *Very similar numbers in the full-deck dataset, and in a 71k random selection 31 Case-studies exemplify real life challenges and BCD opportunities Case study Core set(s) Objective Bacteriology Target-1 Target-2 None Docking Prior hits, privileged scaffolds, pre-plated diversity Gram-negative antibacterial hits without mammalian cytotoxicity Novel starting points for chemistry optimization Chemical matter for a new MOA Malaria Similarity [example] 32 16

17 Target-1: Novel starting points for chemistry optimization Quality determinant: chemotype novelty Triage: dose-response curve shape, analytical QC, ligand efficiency, dissimilar to known chemotypes Source set Screening sets # Confirmed hits # Quality hits 1,157k BCD: 53k Docking: 51k Color by class A B C D 33 Target-2: Chemical matter for a new MOA Quality determinant: MOA validation Triage: Assay interference; orthogonal biophysical assays Source set Screening sets # Confirmed hits # Quality hits 736k BCD: 19k 81 5 Prior hits: privileged scaffolds: 15k 68 3 pre-plated diversity: 18k 73 2 Color by class A B C D 34 17

18 Malaria: Public domain data set to exemplify the workflow Quality determinant: [generic profile with publicly available attributes] Triage: [comparison with published set of 172 confirmed and crossvalidated hits ] Source set Screening sets # Confirmed hits 306k BCD: 50k 31 Similarity: Complete literature hit list Color by class A B C D 35 We re finding good quality confirmed hits Effective exploration! Case study Core set(s) Objective Screening set / Conf. hits / Quality hits Bacteriology None Gram-negative antibacterial hits without mammalian cytotoxicity 76k / 40 Target-1 Target-2 Docking Prior hits, privileged scaffolds, pre-plated diversity Novel starting points for chemistry optimization 104k / 412 / 10 Chemical matter for a new MOA 52k / 232 / 14 Malaria Similarity [example] 50k /

19 Audience Survey Question ANSWER THE QUESTION ON BLUE SCREEN IN ONE MOMENT 37 Which of the following are representations of Chemical Space? * If your answer differs greatly from the choices above tell us in the chat! Assessing chemical space coverage when applying BCD 38 19

20 BCD is designed for coverage: Clustering and complementation Biased selection meta-node uses the complement function of RDKit Diversity Picker in every round 39 Malaria case study example: The impact of complementation Cluster has 55 compounds: 3 B, 50 C, 2 D Coverage settings: Ceiling ( 0.5* (53) )= 4 All three B compounds get selected Task: Pick one C from set of 50 to complete selection 40 20

21 There are class-c compounds that are close to class-b in property space A: Complement set (similarity) B: 250 MW 400; -1 SlogP 3; ROT 5 & no substructure match C: 200 MW 600; -1 SlogP 5; ROT 10 & substructure Flag D: MW > 0... and they are similarly inactive 41 The 7 true active compounds from this cluster are dissimilar in property space Complement design picks one of these, which also is one of the confirmed hits! 42 21

22 Scaffold-based coverage assessment: Two different scaffold definitions 43 a) Bemis & Murcko, J. Med. Chem. 1996, 39, b) Schuffenhauer et al, J. Chem. Inf. Model. 2007, 47, Scaffold-based coverage assessment: Good diversity & complementation # BM scaffolds in conf. hits b # Conf. hits with level-3 scaffold # Level-3 scaffolds in conf. hits b Compound set # Conf. hits Bacteriology BCD Target-1 Docking (4) (2) BCD (4) (2) Target-2 Pre-plated diversity 73 a Prior hits (1) Privileged scaffolds (1) (1) BCD (1) (2) Malaria Similarity BCD (11) (11) All other confirmed hits (11) (11) a) 72 have a BM scaffold b) In parentheses is number of scaffolds shared with another subset Confirmed hit BM scaffold Level-3 scaffold 44 22

23 2D maps of confirmed hits in chemical space show good complementation Target 1 Target 2 Malaria Target focus core: docking, prior hits, similarity Other core: pre-plated diversity Other core: privileged scaffolds BCD Not selected Optibrium s StarDrop Chemical Space tool (visual clustering approach: t-sne) 45 Biased complement diversity selection for effective exploration of chemical space in hit-finding Every well counts... BCD workflow selects high quality representatives to cover a given chemical space o Increased proportion of non-cytotoxic hits from 31% to 62% in bacteriology o Two target-driven hit-finding campaigns identified quality starting points by complementing knowledge-driven subsets with biased diversity sets 46 23

24 Acknowledgements Co-authors: Gianfranco De Pascale, Susan Fong, Mika Lindvall, Heinz Moser, Keith Pfister, Bob Warne, and Charles Wartchow Complement design: Eric Martin Data & workflow infrastructure: Heather Hogg, Manuel Schwarze Antibiotic property space: Folkert Reck Chemical space: Peter Ertl, Ansgar Schuffenhauer Reviewers of the paper (internal & external) All coworkers past & present for sharing and creating drug discovery quality insights 47 Effective Exploration of Chemical Space in Hit-Finding Johanna Hanneke Jansen Director, Novartis Institutes for BioMedical Research Zoe Cournia Investigator Assistant Professor Level, Biomedical Research Foundation, Academy of Athens and JCIM Associate Editor Slides available now! Recordings are an exclusive ACS member benefit. This ACS Webinar is co-produced with the ACS Division of Medicinal Chemistry, the American Association of Pharmaceutical Scientists, and the Journal of Chemical Information and Modeling 48 24

25 Celebrating 5 years & 50 Drug Discovery Webinars! Upcoming ACS Webinar

26 Women in Computational Chemistry Coming out in May 2019! Platform for us to disseminate seminal works Set inspiring examples of female role models Create a network of peers that will instigate change for the new generation of Women in Science in the field of Computational Chemistry Effective Exploration of Chemical Space in Hit-Finding Johanna Hanneke Jansen Director, Novartis Institutes for BioMedical Research Zoe Cournia Investigator Assistant Professor Level, Biomedical Research Foundation, Academy of Athens and JCIM Associate Editor Slides available now! Recordings are an exclusive ACS member benefit. This ACS Webinar is co-produced with the ACS Division of Medicinal Chemistry, the American Association of Pharmaceutical Scientists, and the Journal of Chemical Information and Modeling 52 26

27 All things around ACS Webinars are benefits! We learn about great research topics, we take advantage of our Membership, and we are being friendly to the environment because we can connect from anywhere. Jans Alzate-Morales Pharmaceutical Chemist, Associate Professor, and Director of Ph.D Program in Sciences Mention in Modeling of Chemical and Biological Systems, Centre for Bioinformatics and Molecular Simulations (CBSM), Universidad de Talca ACS member for 9 years strong! yvonnecmartin@comcast.net Be a featured fan on an upcoming webinar! Write to acswebinars@acs.org

28 Join the Division Today! For $25 ($10 for students), You Will Receive: A free digital copy of our annual medicinal chemistry review volume (over 680 pages, $160 retail price) Abstracts of MEDI programming at national meetings Access to student travel grants and fellowships Find out more about the ACS MEDI Division! @AmerChemSociety Contact ACS Webinars at acswebinars@acs.org 56 28

29 Benefits of ACS Membership Chemical & Engineering News (C&EN) The preeminent weekly digital and print news source. NEW! ACS SciFinder ACS Members receive 25 complimentary SciFinder research activities per year. NEW! ACS Career Navigator Your source for leadership development, professional education, career services, and much more ACS Webinars does not endorse any products or services. The views expressed in this presentation are those of the presenter and do not necessarily reflect the views or policies of the American Chemical Society. Contact ACS Webinars at 58 29

30 Upcoming ACS Webinar