Menu

Motsinger-Reif Lab

NC State University

header photo

Dr. Motsinger-Reif has moved to NIEHS

As of December 2018, Dr. Motsinger-Reif has moved to the National Institute of Environmental Health Sciences (NIEHS). She is currently Chief of the Biostatistics and Computational Biology Branch.

Software

To obtain copies of the software listed below, please contact Alison Motsinger-Reif at aamotsin@ncsu.edu to receive up-to-date versions.
 

This web-based tool package genotype quality control for data from the Affymetrix DMET chip or other pharmacogenomic or candidate gene chips, including appropriate quality control for non biallelic markers.

To use this software, please visit our website: http://pgxclean.com


MAGWAS: Multivariate Ancova Genome-Wide Association Software 

Continuing advances in genotyping technology have dramatically increased the amount of data available for genome-wide association studies (GWAS). In many of these studies, multiple distinct responses are observed for each individual. These responses could be measurements over time or space, or could represent related but non-identical outcomes (such as responses to various drug concentrations). One widely used method for analyzing such data are linear models adapted for multivariate responses, such as multivariate analysis of covariance. 

Currently, most popular statistical software packages do not include options for analyzing multivariate outcomes.  MAGWAS was developed to address that need, while still using popular data formats for easy compatibility. Its purpose, as the name indicates, is to provide analysis tools for association studies of single nucleotide polymorphisms (SNP) in genetic data having multivariate outcome and, possibly, multiple covariates. 

MAGWAS tests for the the significance of each locus in a GWAS, where each individual has a vector of related outcomes. For example, each individual could have a response to a medication at three fixed time points, or each tumor cell line could give a response to a medication at four different concentrations.  MAGWAS models the vector of responses for each individual jointly using a well-established multivariate analysis of covariance design.

To obtain MAGWAS, please visit our repository: https://code.google.com/p/magwas/downloads/list

References: 

[1] Brown, Chad C., and Motsinger-Reif, Alison A.(2012) "Software for Genome-Wide Association Studies having Multivariate Responses: Introducing MAGWAS," North Carolina State University Mimeo Serie #2641.  February 16, 2012.

[2] Brown CC, Havener TM, Medina MW, Krauss RM, McLeod HL, Motsinger-Reif AA. Multivariate methods and software for association mapping in dose-response genome-wide association studies. BioData Min. 2012 Dec 12;5(1):21. 


Adaptive Permutation Testing

This code implements an adaptive permutation testing approach for genome-wide association studies.

To obtain this software, please download the R code: choose sampling parameters for adaptive permutation-1.r


GENN: Grammatical Evolution Neural Networks 

Grammatical Evolution (GE) is an evolutionary algorithm inspired by the biological processes of transcription and translation.  Linear genomes, and a grammar (a set of translation rules) are used to automatically evolve optimal solutions to a given problem.  Details of GE can be found at http://grammatical-evolution.org/.  Grammatical Evolution Neural Networks (GENN) utilizes GE to optimize the input, architecture, and weights of neural networks (NN) to detect gene-gene interactions in large scale genomic data. The GENN software in currently in the beta-testing phase.  Investigators who wish to use the GENN and serve as beta-testers should email aamotsin@ncsu.edu.  

The current distribution is downloadable through the following link: genn-gedt-dist_04Nov2010.tgz
 
Reference:
Motsinger-Reif AA, Dudek SM, Hahn LW, and Ritchie MD. Comparison of approaches for machine-learning optimization of neural networks for detecting gene-gene interactions in genetic epidemiology. Genetic Epidemiology 2008 May;32(4):325-40.


Multifactor Dimensionality Reduction (R Code)

Multifactor Dimensionality Reduction is a data-mining tool designed to detect gene-gene and gene-environment interactions by performing an exhaustive search of all loci and performing attribute construction to build high-risk/low-risk models of disease.  MDR has been implemented in a variety of platforms (see Jason Moore’s epistasis.org site for more details and links to other software), and we are distributing an R language version.  

To request a version of the R package MDR.R, please see the CRAN website (http://cran.r-project.org/) or email your request to motsinger@stat.ncsu.edu or sjwood@ncsu.edu.

Reference:
Winham SJ, Motsinger-Reif AA. An R package implementation of Multifactor Dimensionality Reduction. BMC BioDataMining. 2011 Aug 16;4(1):24.


EADRM: Evolutionary Algorithm Dose Response Modeling

EADRM is an evolutionary algorithm approach to nonlinear dose-response modeling.  This approach avoids model assumptions by evolving the best-fit model for the data, as well as performs nonlinear modeling by deriving parameter estimates.  Versions of EADRM are available in both R and JAVA.  

To request a version of the software, please send an email request to motsinger@stat.ncsu.edu or to beam.andrew@gmail.com.

Reference:
Beam A, Motsinger-Reif AA. Optimization of Nonlinear Dose- and Concentration-Response Models Utilizing Evolutionary Computation. Dose Response 2011; 9(3):387-409.


GEDT: Grammatical Evolution Decision Trees 

Grammatical evolution, as mentioned above with GENN, is a flexible evolutionary computing algorithm that can be used to optimize any classification model.  In addition to the NN application described above, we also have a version to optimize decision trees for classification in high-dimensional genetic data. The GEDT software in currently in the beta-testing phase.  Investigators who wish to use the GENN and serve as beta-testers should email motsinger@stat.ncsu.edu.  

The current distribution is downloadable through the following link: genn-gedt-dist_04Nov2010.tgz

 

Reference:
Motsinger-Reif AA, Deohdar S, Winham SJ, Hardison NE. Grammatical Evolution Decision Trees for Detecting Gene-Gene Interactions.  BMC BioDataMining. 2010 Nov 18:3(1):8.

The simulated data used in the above reference is also available by request through the same email contact.

Contact

Alison Motsinger-Reif, PhD

Associate Professor
Assistant Department Head
Department of Statistics
Bioinformatics Research Center

Director
Bioinformatics Consulting and Service Core

University Faculty Scholar

347 Ricks Hall
1 Lampe Dr.
Campus Box 7566
Raleigh NC 27695

Tel: (919) 515-3574
Fax: (919) 515-7315

email: aamotsin@ncsu.edu