MBMDR-4.4.1 is a software that is able to detect multiple sets of significant gene-gene and/or gene-environment interactions in relation to a trait of interest, while efficiently controlling type I error rates. It is mainly used to analyse data were the trait is expressed either on a binary or continuous scale, but can also handle a censored trait.

Operating systemArchitectureMBMDR-4.4.1

Here is the manual (a more detailed one will be available soon) and an overview of the evolution of the options:

The software also contain a man-like help. Type "./mbmdr-4.4.1-mac-64bits.out help" in a terminal to display it.

Here is a script showing how to run the parallel workflow of MBMDR-4.4.1 on a cluster: parallelScript.sh

We also propose an R tool called MoG-Plot (Manhattan over Grail plot) to visualize main-effect and interaction results at a glance: MoG-Plot.r

If you have any question, do not hesitate to contact the author of the program by email: f.vanlishout@ulg.ac.be

History of changes

  • MBMDR-4.4.1: contains the new ONTHEFLY correction method for categorical covariates and a new option to perform a two-stage analysis with a discovery and a replication file.
  • MBMDR-4.3.3: solving a bug when correcting for covariates. Adding the possibility to correct for main effects when the dataset is a survival one (based on the cox proportional hazards model).
  • MBMDR-4.3.2: adding the possibility to handle muliple traits (one by one). Adding the possibility to take account of covariates (working with residuals). Creating a new file containing the models.
  • MBMDR-4.3.1: same version as MBMDR-4.3.2, except that there is a bug when handling covariates expressed on a continuous scale, in the case of a trait expressed on a binary one. This version is not available anymore.
  • MBMDR-4.2.2: replacing the speedMAXT algorithm by the much faster gammaMAXT one. As a bonus, gammaMAXT does control the FWER (whereas speedMAXT did not guarantee it). Revising the default options.
  • MBMDR-4.1.0: sovign a bug when combining speedMAXT with the 3D option. Solving a bug when using -v LONG. Allowing to combine -f and 3D.
  • MBMDR-4.0.3: allowing to combine the speedMAXT algorithm with the 3D option. Solving the bug with the -v option. Allowing MDR input format.
  • MBMDR-4.0.1: contains the new STRAT1, STRAT2 and speedMAXT algorithms. The purpose of the first two is to correct for population stratification. The purpose of the speedMAXT is to perform a MAXT-like multiple-testing, but faster (based on a maximum of 1 million non-zero test-statistics per permutation). This version contains a bug with the -v option.
  • MBMDR-3.1.0: first version with a kind of progress bar. This version also solves an important issue with the --3D option.
  • MBMDR-3.0.3: adding code to handle a censored trait and solving a small bug related to the cholesky decomposition (used to correct for main effects). This version is as fast as mbmdr-3.1.0.
  • MBMDR-3.0.2: refactoring the code and adding an option for performing 3-order effects. This version is about 1.5 times faster than mbmdr-2.7.5.
  • MBMDR-2.7.5: first public C++ version of MB-MDR containing Van Lishout's implementation of maxT. This version handles traits expressed on either a binary or a continuous scale. It can perform a main effect analysis or a two-order interaction analysis.
  • FAM-MDR: R-implementation of the MB-MDR methodology. Very slow and not flexible. Can only handle a trait expressed on a binary scale. Note that this software prepares the data by computing residuals that have removed familial relationships as much as possible, before running the MB-MDR methodology. Removing the familial relationships is done using the polygenic function in the GenABEL package. The C++ software does not remove the familial relationships, allowing you to adopt another modeling strategy.

Older versions

We do not recommand using older versions of the software.
VersionMac 64-bitsMac 32-bitsLinux 64-bitsLinux 32-bits

Here is the corresponding documentation:

For FAM-MDR, download the following R-package