README.md

In HannahVMeyer/DrStable: Estimating the stability of dimensionality reduction results

drStable

The analysis of high-dimensional datasets often requires extracting meaningful variables from the data or compress the data into a more tractable number of features. Feature extraction can be achieved by using dimensionality methods.

drStable provides an interface for easy access to 13 different dimensionality reduction methods: DiffusionMap, DRR, ICA, LLE, Isomap, LaplacianEigenmap, MDS, PCA, kPCA, nMDS, tSNE, UMAP and PEER.

drStable introduces a novel stability criterion which provides the user with a tool of estimating the number of low-dimensional features that can be reliably recovered and thus determine the stable dimensions of the low-dimensional space.

Full documentation is available at http://meyer-lab-cshl.github.io/drStable/.

drStable will be useful in any application that requires the selection of a low-dimensional feature representation of the original data. For instance, in the filed of genetic association studies, these features can be used as the response variable to find genetic variants associated with the low-dimensional phenotype representation. Stable features for association testing harbor two advantages: they increase the power for genetic discovery and guarantee reliable association results.

Installation

Currently, drStable is available in alpha-version via:

library(devtools)
install_github("meyer-lab-cshl/drStable")

We are working on the release of a stable version of the package, including improving test coverage, extension of the documentation and use-case vignettes.

A log of version changes can be found here.

HannahVMeyer/DrStable documentation built on Jan. 29, 2021, 11:42 a.m.