2016 Summer Short Course on Causal Discovery from Biomedical Data & Causal Discovery Datathon

Causal Graphical Models
Giant Eagle Auditorium, Baker Hall

Morning (9:00 am – 12:00 pm) – Registration will be open at 8:00 am

• Introduction
  • Overview of causal graphical models
  • Loading Tetrad
  • Causal graphs/interventions
• Building Models
  • Parametric models: Bayes nets, SEM, other (e.g., generalized SEM)
  • Instantiated models: Bayes net, SEM, generalized SEM

Lunch (12:00 – 1:30 pm) – on your own

Afternoon (1:30 – 4:00 pm)

• Estimation and Model Fit
  • Estimation
  • Inference
  • Model fit
• Hands-on Real-Data Examples

Dinner on your own.

Causal Graphical Models
Giant Eagle Auditorium, Baker Hall

Morning – (9am – 12:00 pm) – Search I

• D-Separation
• Model equivalence
• Basic search algorithms
• Hands-on real-data examples

Lunch (12:00 – 1:30 pm) – on your own

Afternoon (1:30 – 4:00 pm)– Breakout Groups

• 1:30 – 1:50 – Introduction to additional software and Bridges supercomputer
• 1:50 – 2:00 – Breakout into workshop groups by interest and data type
• 2:00 – 3:00 – Instructor led group breakout
  • Load data, specify and test a hypothesis
• 3:00 – 4:00 – student free time with breakout instructor and TA(s)

Dinner on your own.

Biomedical Causal Discovery Overview
Giant Eagle Auditorium, Baker Hall

Morning (9:00 am – 12:00 pm) – Search II

• Latent variable search algorithms
• Time series
• Issues involving measurement

Lunch (12:00 – 1:30 pm) – on your own

Afternoon (1:30 – 4:00 pm) – Breakout Groups

• 1:30 – 2:30 – instructor led group breakout
  • Enter background knowledge
  • Perform basic searches
• 2:30 – 4:00 – student free time with breakout instructor and TA(s)

Evening: Short Course Dinner and Poster Session (included with registration)

O’Hara Student Center, 2nd Floor Ballroom, University of Pittsburgh, 4024 O’Hara Street

• 5:30-6:15 pm – Reception and Poster Session
• 6:15-8:00 pm – Dinner with keynote speaker Dr. Gregory Cooper

Case Studies
Giant Eagle Auditorium, Baker Hall

Morning (9am – 12:00 pm) – Biomedical Case Studies

• fMRI (brain functional connectome)
• Cancer genomic drivers
• Lung disease pathways (susceptibility & progression)
• Genetic regulatory network examples

Lunch (12:00 – 1:30 pm) – on your own

Afternoon (1:30 – 4:00 pm) – Breakout groups

• 1:30 – 2:30 – instructor led group breakout
  • Finalize searches
  • Summarize results
  • Create short report
• 2:30 – 4:00 – student free time with breakout instructor and TA(s)

Dinner on your own.

Breakout Group Reports & Discussion
Giant Eagle Auditorium, Baker Hall

Morning (9:00 am – 12:00 pm) – Wrap Up  & Discussion

• Breakout group reports
• Course question-and-answer period, discussion, and wrap-up
• Evaluations

Afternoon: Causal Discovery Datathon Begins at 1:00 pm in the same location
(Giant Eagle Auditorium, Baker Hall)

Causal Discovery Datathon

Jeremy Espino, MD, MS, Director

Immediately following the Short Course, we will hold a Datathon designed to instruct and challenge biomedical researchers on the use and application of causal modeling and discovery tools in a “bring your own data event”.  You will be required to use our CCD software, including our API (Java, R, Python) and command-line interfaces for Fast Greedy Search (FGS) and any other causal discovery algorithms available for use in June.

A panel of data scientists from our CCD Driving Biomedical Project teams will judge the analyses and award prizes based on size and complexity of data, impact with regard to the causal hypotheses generated, and innovation in the use of CCD tools.

You will need to download CCD software and perform preliminary data formatting in advance of the Datathon. If you attend the Short Course, you will have already completed this process. If you are attending only the Datathon, we will send you formatting specifications for your data so that you can prepare them in advance.

Causal Discovery Datathon