version 1.0 (2012-10-01)
It starts as a copy of BLR version 1.3
Major changes, new list interface, models added for continuous, discrete and
censored traits
For continuous traits:
-BayesA
-BayesB
-BayesC-pi
-RKHS(reproducing kernel Hilbert spaces)
For discrete traits
-Binary responses
-Ordinal responses
New functions to read and write plink files in binary format
version 1.0.1
Initial release
version 1.0.2
Some documentation fixes, some comments added to the code
version 1.0.3
Some methods added: summary, residuals, predict, effects
version 1.0.4
Methods added to deal with heterogeneous variances
Bug fixes for BayesB and BayesC-pi. There was a bug that included
markers that should not appear in the model
version 1.0.5
Using BRR_sets (BRR_windows removed)
Saving effects in binary files & computing windows
version 1.0.6
Never released, problems with reverse depencies
version 1.0.7
Removing OpenMP Flags, C code for registring routines added
Update urls
version 1.0.8
Bayesian Ridge Regression now supports (square) triangular matrices and samples effects efficiently
Sampling from truncated normal distribution is performed using the truncnorm library
Fixes in C code for avoiding integer overflow in sampling functions with large datasets
CITATION added
Several fixes for displaying messages when using stop(), cat(), etc.
version 1.0.9
Added several functions, i)BLRCross and ii)BLRXy which implements efficient fitting of models based on summary
statistics X'X and X'y, which is faster when number of individuals (n) is bigger than the number of markers(p).
Added functions to fit multitrait models with arbitrary number of random effects.