BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:IC Talk: Sparse Matrices and High Performance Computing Meet Biolo
 gy
DTSTART:20210818T101500
DTEND:20210818T111500
DTSTAMP:20260406T185228Z
UID:016cb616c670c8d3567fe1104cefdd34c5ed6ae88629ee416f86e871
CATEGORIES:Conferences - Seminars
DESCRIPTION:By: Giulia Guidi - UC Berkeley\n\nAbstract\nRecently\, the ben
 efit of high-performance computing (HPC) for science has grown rapidly\, b
 eyond traditional simulations to data analysis\, for example\, in genomics
 . Given the vast amount of data and computation involved in such applicati
 ons\, they can require the full computational power and memory of institut
 ional or agency-wide HPC systems.\nOne of the most data- and compute-inten
 sive challenges in genomics is de novo genome assembly\, i.e.\, reconstruc
 ting an unknown genome from redundant\, erroneous genomic sequences. Here 
 we introduce the first distributed memory assembler for long-read sequenci
 ng data\, called ELBA. ELBA introduces sparse matrices as the main abstrac
 tion in this context and makes extensive use of sparse linear algebra comp
 utation and probabilistic modeling. ELBA is up to 2x faster on CPU than an
  algorithm based on distributed hash tables\, which are harder to parallel
 ize. ELBA integrates GPU support in the most compute-intensive stages of t
 he pipeline to take advantage of today's HPC heterogeneous hardware.\nTo e
 nsure that the genomics research community and others\, in general\, can b
 enefit from HPC\, the development of distributed algorithms such as ELBA m
 ust be coupled with efforts to make distributed computing more accessible\
 , as traditional HPC resources are typically reserved for specific researc
 h communities and access to resources is limited. To this end\, we conduct
 ed a performance study to investigate the gap between traditional and clou
 d-based HPC. Until 2018\, cloud-based HPC was not an option for most compu
 tational sciences due to the lack of a low-latency network. Our results sh
 ow that this is changing and that cloud-based HPC is proving to be competi
 tive with traditional supercomputing thanks to faster hardware procurement
  cycles and a significant improvement in network performance.\n\nBio\nGiul
 ia is a PhD candidate in Computer Science at UC Berkeley and a graduate re
 search assistant at the Computational Research Division of Lawrence Berkel
 ey National Laboratory advised by Aydın Buluç and Kathy Yelick. Giulia i
 s a 2020 SIGHPC Computational & Data Science Fellow and a member of the PA
 SSION Lab\, the BeBOp Group\, and the Performance and Algorithms Research 
 (PAR) Group. She received her M.Sc. and B.Sc. in Biomedical Engineering fr
 om Politecnico di Milano. Giulia’s research focuses on developing a nove
 l algorithm for de novo assembly of genomes in distributed memory using lo
 ng-read sequencing data as part of the ExaBiome project\, and on how to ma
 ke cloud computing more accessible for high-performance scientific computi
 ng. Giulia is interested in the intersection of High-Performance Computing
  (HPC)\, Computer Systems\, and Computational Biology as enabling technolo
 gies for faster\, high-quality bioinformatics and biomedical research.\n\n
 More information\n 
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR