Submitting Author Name: Armin Rauschenberger
Submitting Author Github Handle: @rauschenberger
Repository: https://github.com/rauschenberger/corila
Submission type: Pre-submission
Language: en
- Paste the full DESCRIPTION file inside a code block below:
Implements lasso regression (Gaussian, logistic, Poisson, and Cox) for features in heterogeneous groups (e.g., where some groups contain subgroups of correlated features), overlapping groups (i.e., where some features belong to two or more groups), and misspecified groups (i.e., where some group labels are incorrect or missing). Involves an initial ridge regression followed by a final adaptive lasso regression. Exchanges information between features according to their assumed adjacencies or their observed correlations to improve selection and predictive performance. Allows for features that can provide information on other features but must not themselves be selected (i.e., privileged information).
Scope
-
Please indicate which category or categories from our package fit policies or statistical package categories this package falls under. (Please check one or more appropriate boxes below):
Data Lifecycle Packages
Statistical Packages
-
Explain how and why the package falls under these categories (briefly, 1-2 sentences). Please note any areas you are unsure of:
This package implements a regression method for high-dimensional data, with a focus on variable selection and interpretation of estimated effects (thus "regression and supervised learning" and not "machine learning").
Yes, standards are documented in the source code of the R functions and in the unit tests.
- Who is the target audience and what are scientific applications of this package?
This package might be useful to researchers modelling a response based on many predictors, because settings with many predictors often involve predictors that are assumed to be structured by groups or observed to be strongly correlated. This package was developed for biostatisticians and bioinformaticians predicting treatment response from molecular data, but it might also be useful in other fields.
I am not aware of any R packages implementing high-dimensional regression with grouped and correlated features allowing for privileged information (i.e., predictors that are only available during model fitting and not at prediction time). This package extends the R package glmnet (to share information among grouped or between correlated predictors and to account for privileged information). Related R packages are grplasso, grpreg, and gglasso (group lasso), multiridge (group ridge), sparsegl (sparse group lasso), grpregOverlap (overlapping group lasso), and pcLasso (principal component lasso). The difference with respect to these packages is that its selects predictors within groups (also done by sparsegl), shares information between predictors based on assumed adjacencies and based on observed correlations, and that it allows not only for primary predictors but also for auxiliary predictors (i.e., predictors that must not be selected by the model but may provide information during model fitting).
Not applicable.
- Any other questions or issues we should be aware of?:
I somehow only manage to run pkgcheck locally (not via GitHub actions). Identified issues: (i) Two functions have more than 50 lines (cv.corila and multiridge), but I think that I need these lines to pass the numerous arguments to internal validation functions and internal helper functions. (ii) The function name simulate_data is duplicated in other packages, but I believe that this is how it should be called.
Use of Generative AI
If so, please describe usage, and include links to any relevant aspects of your repository. See our blog post for background. (Explicit advice is not yet included in our Dev Guide; we are hoping to update very soon, and ask your cooperation and transparency in the meantime.)
Large-language models (mainly Claude Sonnet 4.6 and GPT-5.4) were used for reviewing R code and documentation. They were also used for writing additional unit tests for debugging purposes, but these were not included in the repository.
Submitting Author Name: Armin Rauschenberger
Submitting Author Github Handle: @rauschenberger
Repository: https://github.com/rauschenberger/corila
Submission type: Pre-submission
Language: en
Scope
Please indicate which category or categories from our package fit policies or statistical package categories this package falls under. (Please check one or more appropriate boxes below):
Data Lifecycle Packages
Statistical Packages
Explain how and why the package falls under these categories (briefly, 1-2 sentences). Please note any areas you are unsure of:
This package implements a regression method for high-dimensional data, with a focus on variable selection and interpretation of estimated effects (thus "regression and supervised learning" and not "machine learning").
Yes, standards are documented in the source code of the R functions and in the unit tests.
This package might be useful to researchers modelling a response based on many predictors, because settings with many predictors often involve predictors that are assumed to be structured by groups or observed to be strongly correlated. This package was developed for biostatisticians and bioinformaticians predicting treatment response from molecular data, but it might also be useful in other fields.
I am not aware of any R packages implementing high-dimensional regression with grouped and correlated features allowing for privileged information (i.e., predictors that are only available during model fitting and not at prediction time). This package extends the R package
glmnet(to share information among grouped or between correlated predictors and to account for privileged information). Related R packages aregrplasso,grpreg, andgglasso(group lasso),multiridge(group ridge),sparsegl(sparse group lasso),grpregOverlap(overlapping group lasso), andpcLasso(principal component lasso). The difference with respect to these packages is that its selects predictors within groups (also done bysparsegl), shares information between predictors based on assumed adjacencies and based on observed correlations, and that it allows not only for primary predictors but also for auxiliary predictors (i.e., predictors that must not be selected by the model but may provide information during model fitting).Not applicable.
I somehow only manage to run
pkgchecklocally (not via GitHub actions). Identified issues: (i) Two functions have more than 50 lines (cv.corilaandmultiridge), but I think that I need these lines to pass the numerous arguments to internal validation functions and internal helper functions. (ii) The function namesimulate_datais duplicated in other packages, but I believe that this is how it should be called.Use of Generative AI
If so, please describe usage, and include links to any relevant aspects of your repository. See our blog post for background. (Explicit advice is not yet included in our Dev Guide; we are hoping to update very soon, and ask your cooperation and transparency in the meantime.)
Large-language models (mainly Claude Sonnet 4.6 and GPT-5.4) were used for reviewing R code and documentation. They were also used for writing additional unit tests for debugging purposes, but these were not included in the repository.