multtest: Resamplingbased Multiple Testing Procedures

1
multtest Resampling-based Multiple Testing
Procedures

• Test Statistics T, F (one and two factor),
  linear and coxph regression, plus non-parametric
  analogs of each.
• Type I error rates FWER, FDR, gFWER and TPPFP.
• Null Distribution Estimators permutation and
  bootstrap.
• MTP object a new class for multiple testing
  results from MTP function
• Slots test statistics, estimates, sample size,
  rawp, adjp, cut-off vector, confidence intervals,
  rejection decisions, null distribution, call,
  seed.
• Methods print, summary, plot, as.list, subset.

2
hopach Hierarchal Clustering w/ Bootstrap
Visualization

• Hybrid of partitioning and agglomerative
  clustering methods
• Main clusters (labels, medoids, ordering within
  clusters),
• Hierarchical tree (ordering in the final level).
• Uses Median Split Silhouette (MSS) to automate
• 1. Number of child clusters at each internal node
  (not binary),
• 2. Decision to collapse two nodes (not
  necessarily siblings),
• 3. Stopping rule to identify the main clusters.
• Distance Metrics Euclidean, correlation,
  cosine-angle, abs value of these.

Hybrid Algorithm
root

• Partition
• Order
• Collapse

4
1
2
3
iterate...
main clusters
final level
ordered list ?