Frema

FReMA R-Package Main Page

Description

FReMA (Fractional-resample Model Averaging) is a generalization of Resample Model Averaging. In the FReMA package we incorporate model changes from LLARRMA (LASSO local automatic regularization resample model averaging), to allow for dominance and additive SNP effects. FReMA combines group LASSO shrinkage of locus groups with fractional resample model averaging and multiple imputation, estimating for each SNP the probability that it would be included in a multi-SNP model in alternative resamples of the data. Specifically, FReMA runs the LLARRMA-rdawg procedure (as well as most of the other variations) described in Sabourin et al. (2015).

For constructive comments and suggestions email the authors
William Valdar and Jeremy Sabourin.

Contents

  1. Updates
  2. Some current limitations
  3. Requirements and installation
  4. Input file
  5. To do list
  6. References

Updates

  • 2015-03-11 FReMA v1.011 uploaded. This update contains fixes for the warnings when installing the package, and corresponding fixes to the documentation.
  • 2014-12-01 FReMA v1.01 uploaded.

Some current limitations

FReMA is limited to linear and logistic regression models, and currently does not feature the ability include unpenalized covariates. If you have know covariates, you may first regress out their effects and run FreMA on the residuals. We do not recommend added covariates as a marker predictor and allow it to be penalized as their scale may effect the local lambda selection in a manner that makes the marker FReMIPs non-informative.

FReMA currently requires packages “glmnet” for fitting lasso based regression paths, and “grplasso” for fitting group lasso regression paths, which are not monitored by anyone on the FreMA project. If changes to this package are made, issues may arise in FreMA. If such an issue does arise, please email William Valdar and we will do our best to fix the issue within our code to account for their changes.

Requirements and Installation

Installing R package FReMA

The R package R/FReMA currently requires R packages “glmnet”, “grplasso”, “matrix”, and “lattice”,as well as the R software.

Installation is supported through R CMD INSTALL terminal interface outside of the RGui.

Please attempt to install “R/FReMA” using the directions below, using Unix/Mac Terminal or Windows Command Prompt through R CMD INSTALL method before contacting Valdarlab with issues regarding installation problems for unsupported methods.

Terminal installation directions follow

    1. Download the latest version of FreMA from here.
    2. Unix (and all users with full R-install capabilities RTools/RFortran installed for Windows/Mac machines) users Open a UNIX shell and go to the directory containing the tar.gz file
    3. Type R CMD INSTALL --clean FReMA_XXX.tar.gz for XXX current package version
  • Installation through RApp Gui is unsupported.

Input files

More details for the .ped file format to be added soon. see package documentation for more details currently.

Command line options

R/FReMA cannot be run from the command line.

To do list

Suggestions to William Valdar and Jeremy Sabourin. Here is the current and incomplete list (in no particular order):

  • Unpenalized covariates within FreMA. For now, a fix is to perform the analysis on residuals after regressing out covariate effects.
  • Adding multiple imputation support for all FReMA options.

References

Sabourin J, Nobel A, Valdar W (2015) Fine-Mapping Additive and Dominant SNP effects using group-LASSO and Fractional Resample Model Averaging. Genetic Epidemiology 39(2):77-88 PMID:25417853

Valdar W*, Sabourin J*, Nobel A, Holmes C (2012) Reprioritizing genetic associations in hit regions using LASSO-based resample model averaging. Genetic Epidemiology 36(5):451-462 PMID:22549815