v1.4.1-alpha

• Drop pmap in favor of ThreadPools.@qthreads
• Add some multivariate IHT documentation
• fix unit tests

v1.4.0

This release contains breaking changes from v1.3.x, mainly in how wrapper functions iht() and cross_validation() are called. The main purpose of this release is to set up our code for our next publication. The next release, v1.4.1, will likely be the finalized code for our 2nd paper.

In summary, we added the following in the v1.4 release:

• Better wrapper functions: now there are only 1 iht() and 1 cross_validate() routine to call, instead of 3. This should improve user experiences.
• Dramatically improve cross-validation efficiency: no more SnpArray copying!
• Dramatically improve efficiency for multivariate analysis
• Add proportion of variance explained (PVE) estimation
• CI testing in Julia 1.6

v1.3.3

This release contains mainly bug fixes and no breaking changes. A prototype code for multivariate IHT is also included, but it has not been well optimized nor is it guaranteed to work.

v1.3.2

Minor tweaks to README and documentation since v1.3.1. No breaking changes nor bug fixes.

v1.3.1

Rename fit to fit_iht so documentation displays properly.

v1.3.0

This version contains breaking changes from v1.2. The main differences are:

• Combine all L0_reg to a single fit function
• Use SnpLinAlg internally for compressed genotype-vector multiplication (this step should be at least 5x faster than v1.2)
• Add a wrapper function so most users can easily run IHT with a single command
• Print informative intermediate results
• MendelIHT now supports custom AbstractMatrix{T} as inputs to run IHT, thus supporting VCF files.
• Fix possible race condition in cross validation.
• Greatly enhance documentation quality

These features are mostly implemented in PRs #11 and #12

Code accompanying paper

Paper was published in GigaScience on June 3, 2020.

IHT for GLM

Works for Normal, Bernoulli, Poisson, and Negative Binomial models. Tested on Linus and Mac using Julia 1.0, 1.1, and nightly.

Julia v0.6 final version

This is the final version of Julia v0.6 code. Future development will target Julia v0.7 and v1.0.