We are thrilled to introduce a host of significant updates and new features in this latest release of sccomp. These enhancements are designed to provide you with more powerful tools for compositional data analysis, improve usability, and offer greater flexibility in your workflows.
1. Support for Random Effects Modeling
One of the most substantial additions is the implementation of random effects modeling within the sccomp framework. This feature allows you to incorporate hierarchical or nested data structures into your analyses, which is particularly beneficial when dealing with complex experimental designs.
Key Advantages:
- Hierarchical Data Analysis: You can now model data that has multiple levels of variability, such as measurements nested within subjects or samples collected across different time points.
- Flexibility in Model Specification: The inclusion of random effects provides greater flexibility in specifying models that accurately reflect the underlying structure of your data.
2. Direct Input of Proportion Data
We have introduced the ability to input proportion data directly into the sccomp functions. This should not be used if counts are present. It is though to model proportions when counts are not available, for example as result of deconvolution.
Key Advantages:
- Greater Data Compatibility: Allows for the integration of data from different sources that may already be in proportion form.
- Enhanced Flexibility: Facilitates the analysis of data types where counts are not available, such as percentages or fractions.
3. Refactoring and Performance Improvements
Significant effort has been put into refactoring the codebase and optimizing performance. This includes rebasing the master branch and cleaning up the code to enhance readability and maintainability.
Key Enhancements:
- Codebase Streamlining: Multiple rebasing efforts (#45, #54, #125, etc.) have resulted in a cleaner, more efficient codebase.
- Model Function Improvements: Refactoring of model functions (#150, #152) enhances computational efficiency and eases future development.
- Nested Grouping with Cmdstanr: Integration of nested grouping capabilities using cmdstanr (#151, #153) allows for more sophisticated statistical modeling.
4. Enhanced Customization and Control
We have added features that give you more control over the analysis process and outputs.
Key Enhancements:
- Custom Output Samples for Variational Bayes: You can now specify the number of output samples when using variational Bayes methods (#137), allowing you to balance between computational speed and estimation precision.
- Deprecation of Redundant Arguments: Cleaning up the function arguments (#155) makes the functions easier to use and reduces confusion.
- Residual Calculation Updates: Changes to how residuals are calculated (#124) improve the accuracy of model diagnostics.
5. Documentation and Usability Improvements
We recognize the importance of clear documentation and have made substantial updates to enhance your user experience.
Key Enhancements:
- Updated README and Vignettes: The README file and accompanying vignettes have been thoroughly updated (#141) to reflect all new features and provide detailed guidance on how to use them.
- Attribute Passing Improvements: Modifications to how attributes are passed between functions (#140) improve the consistency and reliability of the package.
- User Messages and Warnings: Informative messages have been added (#148) to help you understand the progress of computations and alert you to potential issues.
6. Additional Features and Fixes
Several other enhancements and bug fixes have been implemented to improve the overall functionality of sccomp.
Key Enhancements:
- Proportion Difference Calculation: A new feature to calculate the difference in proportions directly (#147), aiding in the interpretation of results.
- Environment Handling in Formulas: Adjustments to formula handling (#142) prevent potential errors related to variable scope and environment.
- Instantiation and Initialization Improvements: Enhancements to how models are instantiated (#136) lead to more stable and faster model fitting.
Full Changelog: https://github.com/MangiolaLaboratory/sccomp/compare/v1.7.12…v1.9.1
For a comprehensive overview of all changes and detailed instructions on how to utilize the new features, please refer to the README.
We believe these updates will significantly enhance your data analysis capabilities using sccomp. The support for random effects modeling and direct proportion data input, in particular, open up new avenues for sophisticated and flexible analyses. We are committed to continuous improvement and welcome any feedback you may have.
Thank you for your continued support, and we hope you find these new features valuable in your research.
What's Changed PR list
- rebase master by @stemangiola in #45
- rebase by @stemangiola in #54
- rebase by @stemangiola in #125
- rebase master by @stemangiola in #128
- rebase by @stemangiola in #131
- rebase by @stemangiola in #132
- Instantiate by @stemangiola in #136
- allow custom output samples for vb by @stemangiola in #137
- update functions to be exposed by @stemangiola in #139
- pass the attribute by @stemangiola in #140
- update README and vignette by @stemangiola in #141
- rebase by @stemangiola in #143
- drop environment from formula and quotes by @stemangiola in #142
- rebase by @stemangiola in #144
- Calculate proportion difference by @stemangiola in #147
- add message by @stemangiola in #148
- Refactor model functions by @stemangiola in #150
- Cmdstanr nested grouping by @stemangiola in #151
- Refactor model functions by @stemangiola in #152
- Cmdstanr nested grouping by @stemangiola in #153
- Cmdstanr by @stemangiola in #84
- deprecate argument by @stemangiola in #155
- Change residuals by @stemangiola in #124
- Proportions by @stemangiola in #126
Full Changelog: v1.7.12...v1.9.1
