Module 3: How to make valid prognostic models with gene expression signatures?

• Title: "How to make valid prognostic models with gene expression signatures?"

• Paper to be examined: “Gene Expression Profiling for Survival Prediction in Pediatric Rhabdomyosarcomas: A Report From the Children's Oncology Group”, J Clin Oncol. 2010 Mar 1;28(7):1240-6 (2010)[1]

• Key claim of the paper: "Metagenes to discriminate patients with good prognosis from those with poor prognosis, with the potential to direct risk-adapted therapy."

• Data and Code

• Schedule:

• • H1: General introduction to the paper/motivation
  • H2-3: Write code to import the data and start computing "meta-genes"

• • H4-6: Aim to fit a predictive model for clinical outcome based on meta-genes

• • H7: Discussion: “Is a model that fits the data necessarily a good predictive model?”
  • H8: Sketch cross-validation approach
  • H9: Summarize results (e.g. on this wiki)

• Key bioinformatics concept of this module:
  • Prognostic models
  • cross validation

• back to UNIL MSc course: "Case studies in bioinformatics 2015"