OffNote AI Talks: Drug Design, Discovery for SARS-CoV2 with ML/Physics models
Drug Design, Discovery for SARS-CoV2 with Machine Learning and Physics-based models
Speaker: Dr. Arvind Ramanathan, Argonne National Labs
Host: Nishant Sinha, OffNote Labs
===
The OffNote Labs AI Talk Series brings you industry experts, researchers and practitioners, passionate to share their learnings and experience -- on innovating and building cutting-edge AI technology / systems which touch and influence the lives of billions of people on this planet.
Our talks are informal, a blend of traditional presentation / podcast, and the audience very technically engaged.
We take delight in unraveling the experience of research and innovation, and celebrate innovators who 'follow the problem', and make complex technology work in the wild.
==
Follow OffNote Labs on LinkedIn. / offnote
Watch previous talks and Subscribe on Youtube: https://bit.ly/31VTMHH
Sign up to our newsletter: http://offnote.substack.com
Read our research articles: / offnote
Web: https://offnote.co
==
00:00 Introduction
05:06 Overview: Drug Design/ Discovery with ML/ Physics Model
06:40 AI Integrated Drug Discovery for SARS - CoV2
08:40 Introduction to Covid-19 and SARS-COV-2
12:00 SARS-COV-2 Genome and ANL/UC Structures
13:33 Molecular Design
14:40 How to Search Billions of Molecules to Find Drug Candidates
15:20 The COVID’19 data Pipeline: Small Molecule Libraries
17:20 First Release of HPC - AI-based Drug Screening
18:40 Natural Language Processing: Dataset and Code
21:46 Improving Docking and Finding better Ligands that Bind to SARS-COV-2 Proteome
24:53 Why not Dock Every Available Compound?
27:20 ML based virtual screening
29:20 Computational Performance
30:40 ML based Virtual Screening
32:53 Improving Docking and Finding better Ligands that Bind to SARS-COV-2 Proteome
34:53 DeepDriveMD Overview: Interleave Simulations and Analytics Adaptively
37:20 DeepDriveMD: DL driven Adaptive Ensembles MD
38:43 AI-Driven MD is an Order of Magnitude Better than Traditional Sampling
41:06 AI-HPC Infrastructure for COVID-19
43:46 AI for Drug Discovery
46:13 Scaling AI-Driven MD Simulations with Heterogeneous Hardware
58:13 Computational Challenges
59:46 DeepDriveMD : Adversarial Autoencoders for Efficient Analysis of MD Simulaton Datasets
1:00:26 Using Fully Convolutional VAE to identify Conformational States in Spike Protein Simulations
1:01:06 DeepDriveMD : Computational Performance
1:04:26 DeepDriveMD : Effective Scientific Performance
1:06:26 Cerebras CS-1: A 15 RU System for Training and Inference in the Data Center
1:07:33 Yes: We can fold Protein in less than 2 hours
1:08:40 We can bind Nsp 10-16 protein
1:08:53 Impacting SARS-COV-2 Medical Thereapeutics
1:16:53 Acknowledgements : Argonne National Labs
1:17:46 Conclusions
#drugdiscovery #drugdesign #SARS-COV2 #sars-cov2-genome #convolutionalVAE #autoencoders #adversarialautoencoders #heterogeneoushardware #deepdrivemd #argonne #nsp-protein #aidrugdiscovery #moleculesampling #searchbillionmolecules #spikeproteinsimulations #conformationalstates