v 0.1.12 # September 23, 2019
  - Add packages to LinkingTo
  - Fix seg fault error due to int function not returning a value
  - restore configure files

v 0.1.11 
v 0.1.10 # June 4, 2019
  - Make compatible with R 3.6.0

v 0.1.9 # April 5, 2017
  - Fix small bug in p-values from wald().  Make sure Sigma is symmetric

v 0.1.8 # August 26, 2015
  - Add documentation that was dropped from the last version

v 0.1.7 # July 10, 2015
  - Remove formula.tools dependency due to error caused by v1.5.4

v 0.1.6 # January, 29, 2015
  - Fix issue in convertToBinary() where the function would fail if relative file paths were used

v 0.1.5 # September 9, 2014
  - Fix issue checking is.nil for big.matrix for some functions

v 0.1.4 # July 31, 2014
  - In read.fam() / read.tfam(), can now read file with arbitrary extension
  - convertToBinary() now supports sample names (i.e. row names)
  - functions getAlleleFreq(), getMissingCount(), getAlleleVariance(), getMACHrsq() now assign names for each entry
  - Add more information to _alleles file produced by convertToBinary() and simpleAnnotation
  - Fix issue with set_missing_to_mean() when the number of missing values is 1 for all columns.  Use foreach() instead of sapply()

v 0.1.3 # July 23, 2014
  - Improve error message in glmApply() and lrgprApply() when features is not a standard R matrix or a FileBacked big.matrix stored as doubles

v 0.1.2 # July 21, 2014
  - Fix glmApply() and lrgprApply() so that terms argument is used correctly.  Previously there was a bug where it was ignored
  - Update convertToBinary() to give better description of why a files does not fit the specified format

v 0.1.1
  - Update error message for plot() of lrgpr() object
  - Update documentation to say that a big.matrix must be re-attached in each R session

v 0.1.0
  - Update documentation for glmApply2()

v 0.0.9
  - Update dependencies

v 0.0.8
  - Improve documentation
  - convertToBinary() writes file with coding of alleles

v 0.0.7
  - Fixed issue in convertToBinary() so that indels in TPED's are treated correctly
  - Fixed a bug where setting the number of threads to k for lrgprApply() or other functions incidentally restricted the number of threads by all subsequent functional calls

v 0.0.6
  - in criterion.lrgpr() and cv.lrgpr(), restrict rank to be less than ncol(features) and length(ord)

v 0.0.5 
  - Update documentation
  - Fix error in getAlleleFreq, getAlleleVariance that made all returned values 1
  - Add getMACHrsq to evaluate imputational quality using MACH's r^2 metric 

v 0.0.4
  - specify columns of features to include/exclude from analysis glmApply() and lrgprApply()
  - lrgpr() and lrgprApply() now handle low rank SVD produced by svd()
  - lrgpr() is MUCH faster for large sample sizes
  - glmApply2() (experimental) implements faster linear regression
  - Backend changes to processing of binary matrix data on harddrive
  - Faster cv.lrgpr() and criterion.lrgpr() by not computing right singular vectors
    - These return objects now compatible with plot()

v 0.0.3
  - graceful exit from glmApply/lrgprApply when response is empty

v 0.0.2
  - Add option to suppress progress bar
  - Throw error when sub.big.matrix is passed as features.  Only matrix and big.matrix are supported