BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260421T090514Z
LOCATION:Bldg. 6 - Room 002
DTSTART;TZID=Europe/Stockholm:20260629T143000
DTEND;TZID=Europe/Stockholm:20260629T150000
UID:submissions.pasc-conference.org_PASC26_sess102_msa117@linklings.com
SUMMARY:IMPECCABLE: An Integrated Drug Discovery Workflow Comprising AI, P
 hysics and Experiment
DESCRIPTION:Peter Coveney (University College London)\n\nI describe the IM
 PECCABLE workflow which runs on Frontier, the world’s first exascale compu
 ter. It implements and couples AI and physics-based (PB) methods to perfor
 m and accelerate a powerful and reliable drug discovery process. It involv
 es screening, generating and evaluating candidate molecules. In an iterati
 ve generative active learning (GAL) loop, the accurate binding free energi
 es from PB methods guide the GAL to formulate compounds with increasingly 
 optimised properties including higher predicted binding potencies. IMPECCA
 BLE also consists of ML-based models for rapidly screening billions of com
 pounds through surrogate docking and generating optimal binding poses for 
 the top ranked ones. The full workflow choreographs and executes these het
 erogeneous components iteratively in the desired order through both upstre
 am and downstream flow of data and information.\n\nApplying IMPECCABLE to 
 two disease-related protein targets, we demonstrate its capability to iter
 atively produce experimentally synthesisable binders with optimised bindin
 g affinities. We exhibit our ability to successfully generate diverse chem
 ical structures that span unexplored regions of chemical space while impro
 ving binding affinities. The potential hits discovered are further optimiz
 ed via an iterative lead-development cycle involving associated chemical s
 ynthesis and assays. The most potent predicted hit exhibits a binding affi
 nity at the nanomolar level.\n\nDomain: Chemistry and Materials, Climate, 
 Weather, and Earth Sciences, Life Sciences, Computational Methods and Appl
 ied Mathematics\n\nSession Chairs: Fabio Affinito (CINECA), Nur Aiman Fade
 l (ETH Zurich / CSCS), and Filippo Spiga (NVIDIA Inc.)\n\n
END:VEVENT
END:VCALENDAR