Bayesian Association Model for Genomic data with Missing Covariates

User Guide

Input Data files

These files are examples of what the input datafiles should look like. The Y,X,Z and R matrices are in gen.data.16.csv and the prior probabilities for the SNPs are in Zprob.csv. For this scenario, note that n=8, p=3, s=5 and there is one missing SNP in the (1,1) position of Z. Notice that the corresponding probability distribution vector in Zprob.csv is the only one that is different from a uniform prior.

BAMD Version 1.2

This latest version does not use any external Fortran code. It calls the Rblas and Rlapack libraries when linear algebra routines are needed. It is now on CRAN at BAMD R package. Here is a copy of the source package.

BAM Version 3.0 (superseded by BAMD)

The new version does not rely on the external libraries. Instead, it contains all the compiled code it needs. There seems to be no need for an autoconfiguration script if your fortran compilers are set up in the usual manner. I will be submitting this to CRAN soon, so if it passes their checks they should provide the Windows binary.

BAM Version 1.1 (superseded by BAMD)

Package Requirements

The package relies on the BLAS and LAPACK libraries for linear algebra routines and on the gsl library for random number generation. If these are installed system-wide, the autoconfiguration script for the Unix-alike versions should find them and install properly. Otherwise, I recommend removing the autoconfiguration script and editing the Makevars file to point the build to the right locations for the required libraries. For Windows, I have a pre-compiled binary package that I have tested on several machines and seems to work fine.