BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20230124T171524Z
LOCATION:D222
DTSTART;TZID=America/Chicago:20221113T103000
DTEND;TZID=America/Chicago:20221113T104500
UID:submissions.supercomputing.org_SC22_sess432_ws_cafcw120@linklings.com
SUMMARY:GPU-Accelerated Differential Dependency Analysis of Single-Cell Tr
 anscriptomics Data
DESCRIPTION:Workshop\n\nGPU-Accelerated Differential Dependency Analysis o
 f Single-Cell Transcriptomics Data\n\nSpeyer, Dong, Kim\n\nComplex disease
 s such as cancer and neurological disorders require a systemic approach to
  understand underlying causes and identify therapeutic targets to help pat
 ients.  More comprehensive analyses, however, often bring significant comp
 utational challenges.  EDDY (Evaluation of Differential DependencY) is a c
 omputational method to identify rewiring of biological pathways between bi
 ological conditions such as drug responses or subtypes of disease [1].  Th
 rough its probabilistic framework with resampling and permutation, aided b
 y the incorporation of annotated gene sets, EDDY demonstrated superior sen
 sitivity to other methods.  Further development integrated prior knowledge
  into these interrogations [2]. However, the considerable computational co
 st for this statistical rigor limited its application to larger datasets. 
  Fortunately, ample and independent computation coupled with manageable me
 mory footprint positioned EDDY as a strong candidate for graphical process
 ing unit (GPU) implementation.  With custom kernels to decompose the indep
 endence test loop, network construction, network enumeration, and Bayesian
  network scoring to accelerate the computation.  GPU-accelerated EDDY cons
 istently benchmarked at two orders of magnitude in performance enhancement
  [3].  EDDY has been applied to the determination of rewired pathways cont
 rolling differing small molecule responses in cancer cell lines [4]. Furth
 er investigations extended this to pathways associated with pulmonary hype
 rtension [5]. \n\nRecent emergence of single cell transcriptomic and spati
 al transcriptomic data raises additional computational challenges, mainly 
 due to an order of magnitude increase in sample size, compared to bulk cel
 l transcriptomic data, often bringing the number of samples to analyze to 
 hundreds of thousands of cells (samples).  This called for additional opti
 mization of the existing EDDY-GPU codes.  By working with a NVIDIA team th
 rough Princeton Hackathon 2022, we were able to dramatically increase the 
 computational speed of the EDDY-GPU.  New sampling strategies has been imp
 lemented to adjust to samples counts at this scale.  In addition, the late
 st code development phase identified various performance bottlenecks, whic
 h not only improved acceleration but allowed for the incorporation of even
  larger gene sets, such as immune pathways.  Hence, EDDY’s statistical rig
 or can now be brought to bear in the inference of specific diagnostic and 
 treatment strategies for the individual patient, and with an implementatio
 n that allows this data analysis to be run on a physician’s desktop within
  reasonable time.  We will present preliminary results using this newly im
 proved EDDY-GPU with single cell transcriptomic data from cancer, Alzheime
 r’s disease, and pulmonary hypertension.\n\nSession Format: Recorded\n\nRe
 gistration Category: Workshop Reg Pass
END:VEVENT
END:VCALENDAR