Collaborative Approach to Supporting Research & High Performance Computing

Collaborative Approach to Supporting Research & High Performance Computing

• 1.
  A Collaborative Approachto Supporting Research and High Performance Computing Marcy Vana, PhD Senior Support Scientist Bernard Becker Medical Library Washington University in St. Louis vanam@wustl.edu
• 2.
  COLLABORATION BEGINS Researchers approachBernard Becker Medical Library (BBML) with a need for genomic data analysis software that is easy to use (graphical user interface) in an environment with computing resources beyond lab computers/laptops BBML approaches Center for High Performance Computing (CHPC) with idea for collaboration
• 3.
  COLLABORATION BEGINS Installation ofgenomics software on the CHPC cluster (WUSTL Galaxy instance & Partek Flow) - web interface for users - analysis jobs run on the CHPC cluster - user programming experience not needed
• 4.
  COLLABORATION BEGINS Installation ofgenomics software on the CHPC cluster - CHPC maintains software - BBML provides user support and training - BBML facilitates licensing when applicable
• 5.
  OUTCOMES Installation of genomicssoftware on the CHPC cluster - positive feedback from researchers - appreciate BBML & CHPC collaboration to make this type of software available
• 6.
  CHALLENGES Installation of genomicssoftware on the CHPC cluster (WUSTL Galaxy instance) - open source software - aspects of software challenging to maintain
• 7.
  CHALLENGES Installation of genomicssoftware on the CHPC cluster (Partek Flow) - commercial software - part of BBML software licensing program - too expensive for some researchers
• 8.
  COLLABORATION GROWS Researchers approachBBML with an interest in learning R CHPC would like to increase awareness about resources and make it easier for new users to get started on the cluster
• 9.
  COLLABORATION GROWS Develop workshopsto introduce the basics of research computing - assume no experience with cluster computing or programming
• 10.
  COLLABORATION GROWS Develop workshopsto introduce the basics of research computing - Computing 101 (using the CHPC cluster) - R - Python
• 11.
  COLLABORATION GROWS Computing 101workshop series - connecting to a remote system - moving data back and forth - basic Unix - shell scripts & batch scripts - interacting with the queueing system
• 12.
  COLLABORATION GROWS R workshopseries - objects, classes, data structures - basic data manipulation & visualization - packages - RStudio - using R on the CHPC cluster
• 13.
  COLLABORATION GROWS Python workshopseries - variables, data types, functions - basic data manipulation & visualization - Numpy, Pandas, Matplotlib - Jupyter Notebook - using Python on the CHPC cluster
• 14.
  COLLABORATION GROWS Develop workshopsto introduce the basics of research computing - same format for all workshops - lecture (30 minutes) - hands-on exercises (90 minutes)
• 15.
  OUTCOMES Develop workshops tointroduce the basics of research computing - praise for working to meet unmet need - increased awareness - seen in new light - fulfilling for staff
• 16.
  CHALLENGES Develop workshops tointroduce the basics of research computing - experience level of attendees varies widely
• 17.
  BBML CONSIDERATIONS Teaching researchcomputing workshops is in line with BBML mission to support research enterprise on campus CHPC is a great partner – very collaborative, equal teaching role for all involved
• 18.
  BBML CONSIDERATIONS Library handlesmajority of administrative tasks associated with workshops (takes time but provides benefits as well) - marketing - registration/waiting lists - communication with registrants - post-workshop surveys
• 19.
  COLLABORATION GROWS AGAIN Joinforces with the recently created Institute for Informatics (I2) to expand content - genomic data analysis with R/Bioconductor - manipulating/cleaning EHR data for analysis
• 20.
  COLLABORATORS BBML Maze Ndonwi Robert Engeszer PaulSchoening CHPC Malcolm Tobias Xing Huang Institute for Informatics Aditi Gupta Madhurima Kaushal Andrea Krussel Philip Payne