BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20210402T160551Z
LOCATION:Track 8
DTSTART;TZID=America/New_York:20201111T153100
DTEND;TZID=America/New_York:20201111T153500
UID:submissions.supercomputing.org_SC20_sess198_ws_whpc101@linklings.com
SUMMARY:Molecular Design Using GraphINVENT
DESCRIPTION:Workshop\n\nMolecular Design Using GraphINVENT\n\nMercado\n\nG
 raphs are widespread mathematical structures that can be used to describe 
 an assortment of relational information, and are natural choices for descr
 ibing molecular structures. Recently, there has been an increase in the us
 e of graph neural networks (GNNs) for modeling patterns in graph-structure
 d data, including graph-based molecular generation for pharmaceutical drug
  discovery. The guiding principle behind graph-based molecular design can 
 be boiled down to generating graphs which meet all the criteria of desirab
 le drug-like molecules. \n\nWe have applied GNNs to the task of molecular 
 generation and recently published GraphINVENT[1,2], a platform for graph-b
 ased molecular design using message passing neural networks (MPNNs) and a 
 tiered feed-forward network structure to probabilistically generate new mo
 lecules one atom/bond at a time. Graphs as data structures are very memory
 -intensive objects compared to the alternative, string-based approaches fo
 r molecular generation. As such, GraphINVENT was optimized so as to succes
 sfully train on large molecular datasets (e.g. millions of small molecules
 ) using GPUs.\n\nGraphINVENT models can quickly learn the underlying distr
 ibution of properties in training set molecules without any explicit writi
 ng of chemical rules. The proposed models perform well for molecular gener
 ative tasks when benchmarked using the Molecular Sets (MOSES) platform [3]
 . Our work illustrates how deep learning methods can enhance drug design a
 nd shows that graph-based generative models merit further exploration for 
 molecular graph generation.\n\n\nReferences\n\n1. 	Mercado R, Rastemo T, L
 indelöf E, et al. Graph Networks for Molecular Design. 2020. doi:10.26434/
 chemrxiv.12843137.v1\n2. 	Mercado R, Rastemo T, Lindelöf E, et al. Practic
 al Notes on Building Molecular Graph Generative Models. 2020. doi:10.26434
 /chemrxiv.12888383.v1\n3. 	Polykovskiy D, Zhebrak A, Sanchez-Lengeling B, 
 et al. Molecular Sets (MOSES): A Benchmarking Platform for Molecular Gener
 ation Models. 2018. doi:arXiv:1811.12823v1\n\nTag: Education, Training and
  Outreach, Professional Development, Workforce Development\n\nRegistration
  Category: Workshop Reg Pass
END:VEVENT
END:VCALENDAR