Augmented Intelligence in Drug Discovery Xchange
West Coast 2021
San Francisco
September 23rd
Welcome to hubXchange’s West Coast Augmented Intelligence (AI) in Drug Discovery Xchange 2021, bringing together executives from pharma and biotech to address and find solutions to the key issues faced in AI-led drug discovery.
Discussion topics will cover Data Quality, Target Identification, Lead Generation, Lead Optimization and Drug Response Prediction.
Take advantage of this unique highly interactive meeting format designed for maximum engagement, collaboration and networking with your peers.
Data Quality
- Industry perspective from a vendor
- Predicting the impact of AI/ML on drug discovery
- Our recommendation for the ecosystem
Chief Solutions Officer
Excelra
Haranath Gnana has over two decades of strategy consulting and analytics-led solutioning experience across Life Sciences & Healthcare. His expertise involves solving complex business problems with data & analytics and enabling enterprises to become data-driven organizations. Before joining Excelra, Haranath co-founded two start-ups in the Bay Area where he extensively worked on AI/ML technologies. He has also worked for leading companies such as Kaiser, Genentech, Franklin Templeton, Cisco, and TiVo.
- Explore the effective, and realistic approaches to annotate and share data; define data quality aspects for ML applications
- How do we best utilize and interpret data of quality?
- Discuss why executing data quality is a continuous process
- Share experiences in tackling data quality challenges
Associate Director Protein Engineering
Bristol Myers Squibb
Sean West is a computational protein engineer at Bristol Myers Squibb. He has spent over 15 years using computational methods to understand and explore protein interactions, from transcription factors and RNA binding proteins, to antibody-antigen recognition. Throughout the antibody discovery process, we use a variety of methods to produce molecules with good specificity, affinity, immunogenicity, and biophysical characteristics. Computational workflows can be used to supplement experimental techniques but require good datasets. Part of the challenge of antibody discovery and molecular design is generating datasets appropriate for computational analyses, a challenge faced by many in our industry.
Discuss the Challenges in Acquiring and Leveraging Clean Data in Drug Discovery
- What are the areas in drug discovery where clean data is most lacking?
- How do we harmonize disparate data formats from different sources?
- How can data be stored and organized for maximum utility AND security?
CEO
Collaborative Drug Discovery
Barry Bunin, Ph.D. is CEO of Collaborative Drug Discovery, Inc., provider of the CDD Vault research informatics platform and BioHarmony drug data store. Prior to CDD, he was an Entrepreneur in Residence with Eli Lilly & Co., and was also the founding CEO, President, & CSO of Libraria (now Eidogen-Sertanty). Dr. Bunin is on a patent for Carfilzomib, a protease inhibitor drug for the treatment of multiple myeloma. Dr. Bunin received his B.A. from Columbia University and his Ph.D. from UC Berkeley, and has co-authored textbooks on cheminformatics and chemical synthesis.
Networking Lunch
- Share experiences in multi-omics data acquisition and curation.
- Discuss sources of bias in data, and approaches to overcome these problems.
- What are the best methods to facilitate data integration?
- How do we best leverage multi-omics data for drug discovery?
Head of Computational Biology
Verge Genomics
Victor Hanson-Smith, PhD, leads a world-class team of computational biologists to discover new drugs for neurodegenerative diseases, including ALS and Parkinson’s Disease. His team’s work combines a wide range of methods, including genomics, systems biology, machine learning, and supercomputing. Prior to joining Verge Genomics, Victor completed his PhD in computer science at the University of Oregon, and a post-doctoral fellowship at University of California San Francisco. He has published extensively on topics related to eukaryotic genome function and evolution.
2:35 – 3:05pm
Combining text analytics + semantic enrichment with Elsevier’s content to power AI/ML initiatives
- Generate artifacts using Elsevier’s large repository of full text articles, domain specific scientific databases containing third party publisher information.
- Leverage Elsevier’s industry standard drug discovery ontologies and taxonomies to create knowledge graphs, network maps and machine learning models.
- Transform unusable text to data in a richly annotated, machine-readable and standardized data format
- Harmonize technology through scientific ontologies adhering to public standards
4:10 – 5:10pm
- Discuss how to balance data obfuscation with data utility for meaningful data sharing
- Explore if organisations are over-valuing their data assets and preventing meaningful data sharing
- Explore the ideal composition of data-sharing consortiums and initiatives- larger pharma and biotech versus smaller start-ups
Director, Computational Chemistry
Novartis
Eric Martin has a Ph.D. in physical-organic chemistry from Yale University. He has worked in computational drug design and herbicide design for over 35 years. He is currently developing novel methodologies for two areas of drug discovery:
1) developing “Profile-QSAR”, a massively multitask machine learning method that builds experimental-quality virtual screening models for over 8000 IC50 assays, and
2) “rational oral bioavailability design” during lead optimization by applying machine learning and global sensitivity analysis to physiologically-based pharmacokinetics simulations.
For the former, Eric was awarded the lifetime title of “Novartis Leading Scientist”.
Target Identification
- Industry perspective from a vendor
- Predicting the impact of AI/ML on drug discovery
- Our recommendation for the ecosystem
Chief Solutions Officer
Excelra
Haranath Gnana has over two decades of strategy consulting and analytics-led solutioning experience across Life Sciences & Healthcare. His expertise involves solving complex business problems with data & analytics and enabling enterprises to become data-driven organizations. Before joining Excelra, Haranath co-founded two start-ups in the Bay Area where he extensively worked on AI/ML technologies. He has also worked for leading companies such as Kaiser, Genentech, Franklin Templeton, Cisco, and TiVo.
- How should public data be manipulated to expedite drug target selection; identification of patient demographics for clinical trials; and spotting of trends and novel technologies
- What public domain sources should we access and what are the right sources to extract
- Extracting the information component- what to parse and what tools are available?
- How to transform extracted information into actionable knowledge
- Discuss the challenges of bias; balancing novelty vs. well-established facts; and sparse data and over-generalization
Director of Statistical Genetics
BioMarin Pharmaceuticals
Karol Estrada, PhD, is a strategic, enterprising, and accomplished Human Genomics Executive respected for 15+ years of experience as a leader in human genetics with a focus on integrating genomic knowledge to discover, validate, and improve drug targets. He is the Sr. Director of Statistical Genetics at the Human Genetics department in BioMarin Pharmaceuticals. He is an internally recognized researcher in the field of human genetics with over 60 publications in high impact factor journals.
Before joining BioMarin, he hold increasing levels of responsibility roles at Biogen. He has served on several international consortia focused on the design and conduct of large genetics studies for a range of human traits and diseases.
Dr. Estrada’s teaching in the Master of Science in Bioinformatics focuses on the use of modern approaches for the study of human genetics, including statistical genetics and genetic epidemiology.
Education
M.S., Monterrey Institute of Technology
Ph.D, Erasmus University
Postdoctoral training: Broad Institute of MIT and Harvard
- How do we assess and test the quality of a dataset? What threshold does it need to hit in order to be included in training an AI/ML model?
- How do we verify possible dataset challenges (e.g. outliers) at scale? Is it even possible beyond a handful of data points?
- What do you think about the peer review process to be brought down to the data world? Is there is a parallel in that world, and if not can there be one? Should we be adopting dataset standards that already exist?
Director, Scientific Products
Excelra
Norman comes with a rich product management background that includes experience in developing and launching enterprise solutions. He brings a user centric and analytically driven approach; the ability to identify areas of growth and opportunity; and experience in setting strategic goals that align with the business objectives of the organization. Norman is leading Excelra’s rapid foray into technology solutions to accelerate innovation in drug development. He holds a Master of Liberal Arts, Extension Studies with a focus in Biology, from Harvard University.
Founder and Chief Executive Officer
Glamorous AI
Noor is a serial biotech entrepreneur with a track record of achievements in AI having held an Assistant Professorship from Aalborg University. Noor has published numerous papers in the field of artificial intelligence and its application to drug discovery and is an inventor on a handful of patents. She is passionate about data and AI and on a mission to cure disease with the power of human and machine learning. She is currently the CEO at GlamorousAI, a biotech company that pushes the boundaries to what is possible with AI to cure debilitating diseases. She is a recognised healthcare leader, MIT innovator under 35 and in BBC 100 women for 2019.
12:50 – 1:50pm
- How accurate are current models for disease pathways- can we confidently predict the outcomes of pathway perturbation?
- Discuss how an understanding of disease biology will serve as the main component to AI- allowing the model to highlight drug targets for R&D focus
- Explore the communications gaps between domain experts and data scientists to ensure that data collection and AI model selection are fit-for-purpose
- Discuss the current landscape of knowledge graph application in pathway modelling
Senior Director, Immunology Discovery
Trex Bio
Ali Zarrin obtained his Ph.D. from University of Toronto in Immunology and then did his postdoctoral study with Fred Alt at Harvard University. He joined Genentech in August 2007 where he led several initiatives to screen and advance multiple therapeutic programs covering diverse pathways in inflammatory and autoimmune indications. Since August 2019, he is currently the senior director of discovery in TRex Bio responsible for target discovery and preclinical therapeutic development in cancer, inflammatory and autoimmune diseases.
4:10 – 5:10pm
- How can AI/ML be used to develop drug target selection and exclusion criteria?
- Explore effective approaches in using AI to expand understanding and build confidence in targets
- True artificial intelligence-led target selection and triage with minimal human input- are we there yet?
Director, Head of Computational Chemistry,
Sutro Biopharma
Valery Polyakov is a Director of Computational Chemistry at SUTRO Biopharma. He works in the field of AI and Machine Learning leveraging Big Data to facilitate drug discovery. He also oversees computational infrastructure development for the Department of Chemistry. Prior to that, he worked at Novartis, developing new drugs for treating cancer and infectious diseases. A compound that he co-invented is now in Phase II of clinical trials. Before that, he worked for Sanofi in molecular modeling and chemoinformatics groups. He holds a Ph.D. in Chemistry from Kiev Tshevchenko University (Ukraine) and prestigious awards including the Fulbright Scholarship. Valery authored more than 30 published papers and 18 patents and patent applications.
Lead Generation
- Industry perspective from a vendor
- Predicting the impact of AI/ML on drug discovery
- Our recommendation for the ecosystem
Chief Solutions Officer
Excelra
Haranath Gnana has over two decades of strategy consulting and analytics-led solutioning experience across Life Sciences & Healthcare. His expertise involves solving complex business problems with data & analytics and enabling enterprises to become data-driven organizations. Before joining Excelra, Haranath co-founded two start-ups in the Bay Area where he extensively worked on AI/ML technologies. He has also worked for leading companies such as Kaiser, Genentech, Franklin Templeton, Cisco, and TiVo.
- Share obfuscated TPPs and the tactics of AI/ML application for developing the TPPs
- Explore other AI tools or omics approach from different therapeutic areas to light a new path for your TPP development
- What is the ideal combination of human interaction and AI for the development of TPPs?
Head, West Coast and Cananda Search and Evaluation
Novartis
Dr. Leeds is responsible for identifying new technology from organizations in the western US and across Canada for potential partnering with Novartis, and potential partners for out-licensing NVS assets. Jennifer joined Novartis in 2003, co-invented and co-led the project team for the novel antibacterial LFF571 through Ph2, and became Head of Antibacterial Discovery in 2010. In 2018, Jennifer joined Novartis BD&L, supporting out-licensing of anti-infective programs to Boston Pharmaceuticals, Gilead, and Amplyx Pharmaceuticals and in-licensing deal flow across all Novartis disease areas. Jennifer is on the Novo Holdings REPAIR Impact Fund SSB, is an Expert Reviewer for Innosuisse, a member of imYEG Council of Founders, a mentor for California Life Sciences Institute start-ups and for Nex Cubed, Y Combinator, and IndieBio portfolio companies. Jennifer did her postdoc at Harvard Medical School, received her BSc from Cornell University and her PhD from the University of Wisconsin-Madison.
No session
- Discuss how AI can be effectively applied for the development of lead generation models using internal and external data
- Explore the current AI capabilities in predicting drug-like properties
- How much should you trust your model? How can you mitigate and plan for uncertainties in the model?
Senior Director, Structural Chemistry, Gilead
Gilead
Bill received his Ph.D. from Wesleyan University in Middletown, CT, for his work in ab initio method development. He subsequently worked with Roald Hoffmann at Cornell where he applied semi-empirical methods to the design of novel materials. He then joined Charles L. Brooks III at Scripps in San Diego to use molecular dynamics to understand the determinates of protein structure. Subsequently, he was employed at several San Diego & Bay Area biotechnology and small pharmaceutical companies (e.g. CombiChem, Neurocrine) as a computational chemist or as a cheminformatician.
After a few years of increasing his software knowledge (as a solution architect and product manager at Biovia), he returned to chemical informatics with a data science focus at Novartis in Emeryville. After the site’s transition away from therapeutic research for virology and bacteriology, he was asked to join Gilead. At Gilead, he leads the Research Informatics group and is responsible for data systems, research workflow systems and predictive chemical informatics.
2:35 – 3:05pm
Poster Session: The identification of novel therapeutic hits to estrogen receptor α using machine learning on DNA-encoded
library screening data
- DNA-encoded library technology at X-Chem
- Case study: machine learning on ERα screening data
- X-Chem’s machine learning offering and future directions
4:10 – 5:10pm
- How can AI/ML be applied for the development and selection of screening libraries and assays
- Explore the best AI approaches to improving virtual screening outcome and Hit ID
- Discuss how to leverage AI/ML to identify leads with in-vivo safety and efficacy
Director, Computational & Structural Chemistry
Merck
Alan Cheng is a Director at the Merck & Co. research laboratories in South San Francisco, where he manages a group working on the discovery of small molecule and biologics therapeutics using biophysical modeling, informatics, and machine learning approaches. Previously, he was at Amgen for 11 years, and Pfizer for 5 years. He has contributed to over five clinical molecules and is an author or co-inventor on over 70 peer-reviewed publications and patents. He received his Ph.D. at UCSF in Biophysics, and his undergraduate degrees at UC Berkeley in Electrical Engineering & Computer Science, and Molecular and Cellular Biology. He also currently chairs the Bioinformatics M.S. program at Brandeis University.
Lead Optimization
- Industry perspective from a vendor
- Predicting the impact of AI/ML on drug discovery
- Our recommendation for the ecosystem
Chief Solutions Officer
Excelra
Haranath Gnana has over two decades of strategy consulting and analytics-led solutioning experience across Life Sciences & Healthcare. His expertise involves solving complex business problems with data & analytics and enabling enterprises to become data-driven organizations. Before joining Excelra, Haranath co-founded two start-ups in the Bay Area where he extensively worked on AI/ML technologies. He has also worked for leading companies such as Kaiser, Genentech, Franklin Templeton, Cisco, and TiVo.
9:05 – 10:05am
- Explore application of AI in speeding drug-design-test-analysis cycle
- What are the current successes and challenges in AI-led small molecules optimization?
- How effective are AI at suggesting novel, synthesizable compound design?
- Thinking of ADMET at the stage of compound design
- Retrosynthesis to suggest more efficient path to make compounds
Staff Research Scientist,
Google
JW Feng is a Staff Research Scientist and leads the drug discovery team in Google Applied Science. He is applying Google technologies, including deep learning, to accelerate the discovery of small molecule drugs. JW was an early employee in a biotech startup, Denali Therapeutics, where he built the molecular modeling and data science group to support both small molecule and biotherapeutics discovery. Key contributions from JW lead to the invention of multiple molecules entering clinical trials. Prior to Denali, JW was a Scientist at Genentech supporting small molecule drug discovery. He is an experienced drug hunter who co-invented >15 leads that progressed into IND-enabling studies. Six of those candidates advanced to clinical trials. JW received a PhD in Computational Biology at Washington University in St. Louis and bachelor of science degrees in Computer Science and Biochemistry at The Ohio State University.
11:50am – 12:50pm
Network Lunch
12:50 – 1:50pm
AI/ML Applications in Macromolecules Optimization
- How can AI/ML be applied for the optimisation of macromolecules?
- Explore the potential of AI/ML in the characterisation and mapping of immunogenicity
- Discuss where AI/ML can be applied to streamline macromolecules production
- All predictive models have uncertainties- how to you search and account for them to de-risk clinical translation?
Executive Director Research – Data Analytics & Attribute Sciences
Amgen
Dr. Peter Grandsard is currently an Executive Director of Research at Amgen responsible for the characterization of pre-clinical therapeutic candidates and for the advancement and implementation of data and computational sciences in Research. Trained as a chemical engineer (BE/ME) and as an analytical chemist (Ph.D.), he started at Amgen 25 years ago, as a scientist designing and implementing new laboratory automation and instrumentation. Later he started leading an analytical function with a remit to analyze therapeutic candidates and reagents, biologics or synthetics alike, in order to understand their structures, their physical-chemical attributes, and their protein target binding properties. Hence it is no surprise Peter also takes interest in in-silico methodologies to understand biological systems or design new therapeutics, hand-in-hand with in-lab data generation.
1:55 – 2:25pm
2:25 – 2:35pm
2:35 – 3:05pm
No session
3:10 – 3:40pm
3:40 – 4:10pm
4:10 – 5:10pm
- In silico assessment of the sequence, as it relates to human and preclinical species, and on-target effects
- Off-target risk assessments
- How to assess chemistry modification using AI
Principal Scientist
Takeda
Heather has over 20 years of experience in the pharmaceutical industry in both large pharma and small biotech spanning multiple modalities, including small molecules, oligonucleotides and antibodies. Her background is in computational biology from Johns Hopkins University and her career has focused on supporting pharmacology and toxicology. Specific to oligonucleotides, she has developed expression signatures for miRNA target modulation used by biologists to assess delivery and target PD, as well as approaches for oligonucleotide target identification, off-target assessment and toxicity risks. She is currently a Discovery Toxicologist at Takeda Pharmaceuticals where she is using bioinformatics to assess target and modality risks.
Drug Response Prediction
- Industry perspective from a vendor
- Predicting the impact of AI/ML on drug discovery
- Our recommendation for the ecosystem
Chief Solutions Officer
Excelra
Haranath Gnana has over two decades of strategy consulting and analytics-led solutioning experience across Life Sciences & Healthcare. His expertise involves solving complex business problems with data & analytics and enabling enterprises to become data-driven organizations. Before joining Excelra, Haranath co-founded two start-ups in the Bay Area where he extensively worked on AI/ML technologies. He has also worked for leading companies such as Kaiser, Genentech, Franklin Templeton, Cisco, and TiVo.
- What is the promise and limitation of current approaches to simulate ADMET in silico?
- Which aspects of ADMET can be currently modeled mechanistically vs. empirically vs. neither?
- What types of experiments or data would help to address blind spots in our understanding and ability to predict ADMET?
- How can AI/ML narrow the translation gap between preclinical models and human?
Principal Scientist
Genentech
Dr. Wei is an experienced drug hunter who has a track record of applying computational approaches to solve difficult problems, including ADMET challenges, in drug discovery. At Genentech, he has worked on multiple disease areas (oncology, immunology, neurodegenerative & infectious diseases) and gained expertise in several therapeutic modalities (including antibody-drug conjugates, covalent inhibitors & degraders). He has directly contributed to the discovery of 6 molecules that advanced into clinical trials and is a co-inventor of 18 granted patents. Prior to Genentech, BinQing was a Senior Scientist at Pfizer. He received his PhD at Northwestern University in Evanston, IL and postdoctoral training at Scripps Research Institute in La Jolla, CA.
12:50 – 1:50pm
Achieving Efficacy and Safety In Vivo Using AI- What are We Missing?
- Explore the current gaps and opportunities for computational algorithms to in vivo relevance
- Biological data is often driven by feasibility/technology push, and not by need/scientific pull – how do we align both?
- Discuss if new types of data and technologies should be evaluated independently by individual companies, or collaboratively to de-risked in vivo transition
- Explore the current capabilities of AI in bridging hazard to risk and assays to organ risk assessments
Bioinformatics Scientist Lead, Bioinformatics Lead on Immuno-Oncology Drug Development
Arcus Biosciences
Ning is a bioinformatics scientist at Arcus Biosciences, where he is responsible for drug discovery, target identification, and indication prioritization in multiple immune-oncology drug programs. Notably, Ning pioneered the 3D organoids based drug response prediction technology, which largely facilitated cancer model selection. During his PhD, he studied with Professor Alexander Hoffmann at UCLA, where he built up computational framework to identify synergy between key immune response signaling pathways in macrophage. Ning holds a PhD in Bioinformatics from UCLA.
2:35 – 3:05pm
No session
4:10 – 5:10pm
AI Prediction of Drug Responses in Patients Using Biomarkers and Genomics
- Discuss state-of-the-art machine learning powered approaches and success stories to analyze OMICS biomarkers and redefine patient stratification models
- Explore Real World Evidence (RWE), mining unstructured text, semantic integration and leveraging several publicly available information data assets are powering the next generation biomarker identification
- Understanding limitations and caveats when Using AI/machine learning predictions
Ranga Chandra Gudivada, Director AI Innovation, Strategy & Partnerships, Advanced Analytics & Data
Global Early Innovation Partnering Lead for Data Sciences
Johnson & Johnson
Bevan Emma Huang, Ph.D., is the Global Early Innovation Partnering Lead for Data Sciences at Johnson & Johnson Innovation, where she sources, evaluates, and facilitates external partnering opportunities in data sciences across all sectors of Johnson & Johnson. Previously, she led Population Analytics in Janssen Research & Development from 2015-2018, where her group helped to develop and support precision medicine initiatives through integrating multiple data modalities such as omics, sensors, and electronic health records for risk prediction, target and biomarker discovery. From 2007-2015, Emma held positions at CSIRO in Brisbane, Australia, where she received numerous grants and awards to develop novel methodology for statistical genetics and data integration.