statistical forays of a modern biodata scientist

my left hemisphere imagining motion by Anisha Keshavan

Daniel Himmelstein

September 4, 2015

Online at slides.com/dhimmel/statistical-forays

Original content released under CC0, attribution via hyperlink

MEDLINE Topic Cooccurrence

MEDLINE topics

http://thinklab.com/d/67

  • curators read abstracts and annotate topics
     
  • 21 million articles
  • 5,594 journals
     
  • cooccurence of two topics indicates a relation

Contingency Table

Independence of variables test:

  • Fisher's exact test
  • chi-squared goodness-of-fit
  • Barnard's test

anatomy

symptom

Mining MEDLINE for disease context

meta-analysis

Combining measures of significance:

Combining measures of effect size:

  • weighted average
  • fixed effects model
  • random effects model

LINCS L1000

← genes 

permutation testing

Network edge swaps

machine learning

Say no to models that overfit

Look for:

testing performance ≈ training performance

But still:

evaluate performance based on testing

Regularized regression:

  • prevents overfitting
  • glmnet package in R
  • ridge -- coefficient shrinkage
  • lasso -- coefficient shrinkage & variable selection

There are two cultures in the use of statistical modeling to reach conclusions from data. One assumes that the data are generated by a given stochastic data model. The other uses algorithmic models and treats the data mechanism as unknown. The statistical community has been committed to the almost exclusive use of data models. This commitment has led to irrelevant theory, questionable conclusions, and has kept statisticians from working on a large range of interesting current problems. Algorithmic modeling, both in theory and practice, has developed rapidly in fields outside statistics. It can be used both on large complex data sets and as a more accurate and informative alternative to data modeling on smaller data sets. If our goal as a field is to use data to solve problems, then we need to move away from exclusive dependence on data models and adopt a more diverse set of tools.

Leo Breiman (2001) Statistical Modeling: The Two CulturesStatistical Science

Sandler Neurosciences Center

Sergio

Lab retreat in Tahoe

Northstar California Resort

February 2015

Egle Cekanaviciute 

fascinating discoveries on the MS

microbiome

Made with Slides.com