LUCIDus v2.2.1

1. Update vignette; create package website; update citations.

LUCIDus v2.2

New features

1. New function lucid: an estimation function integrating model and variable selection. We recommend user to use this new function for estimating LUCID models. It calls the workhorse functions est_lucid and tune_lucid in the back-end.

Major changes

1. Rename est.lucid and boot.lucid to est_lucid and boot_lucid.
2. Update vignette accordingly.
3. Use mclust to choose the optimal geometric model for omics data Z by default.

LUCIDus v2.1.5-1

fix error messages in testing.

LUCIDus v2.1.5

New features

1. Add progress bar for boot.lucid function to track how many iterations are done.
2. Add verbose parameter in est.lucid to disable automatic output in R console.

Other changes

1. Fix bug for boot.lucid
2. Fix bug for predict_lucid

LUCIDus v2.1.2

New Features:

1. Use log-sum-exponential trick to avoid over/underflow
2. Implement the GMM with missing data method (Zhang 2021) for estimating the association X->Z
3. Add more checks for data input format
4. Add new option - init_impute for initializing imputation of missing data
5. Add new parameter - seed to conveniently set random seed for EM algorithm

Other Changes:

1. Disable glmnet when estimating the association G->X since glmnet is not compatible with single exposure scenario
2. Delete unused parameters in boot.lucid
3. Deprecate unnecessary parameters in def.lucd

LUCIDus v2.1.1

1. Use Ridge regression to estimate G to X effect in order to stabilize effect estimate
2. Fix bugs in estimating binary outcome

LUCIDus 2.1.0

New features

A new variable selection framework is applied to LUCID.

• A lasso type penalty is applied on the mean of biomarkers
• A glasso method is applied on the variance-covariance structure to achieve sparsity covariance matrix
• We apply a new variable selection criteria, which takes both mean and coviarnce matrix of biomarkers into account.

Other changes

• Fix bugs in pred.lucid(). Now it can predict both latent cluster and the outcome.

LUCIDus 2.0.0

This is a feature update to the whole package. It rewrite all the codes to make the model fitting procedure much faster (10 to 50 times) than v1.0.0. Also, the grammar of LUCID changed to a more user-friendly version. (Please note, this version is not backward compatible)

New features

• est.lucid(): previously called est_lucid. Fit the LUCID mode much faster; use mclust to initialize and produce a more stable estimate of the model; fix the bugs dealing with missing values in biomarker data.
• summary.lucid(): previously called summary_lucid. An S3 method function which can directly be called by summary; provide with a nice table with detailed interpretation of the model; add option to calculate 95% CI based on result returned by boot.lucid().
• plot.lucid(): previously called plot_lucid. An S3 method function which can be directly called by plot; change the color palette.
• predict.lucid(): previously called pred_lucid. An S3 method function which can be directly called by predict.
• boot.lucid(): previously called boot_lucid. Provide with a neat output.

Other changes

• Update the vignette.
• Updated the citation after getting published by the Bioinformatics;
• Minor bug fixes.

Preparation for CRAN submission

add citation, update vignette, spell checking