 |
New Features
- Compendium
- The paper at http://www.bepress.com/bioconductor/paper2/ describes the idea of a compendium.
- The tables, plots, and other output from HDBSTAT now form a compendium.
- Updated ANOVA analysis action to output preprocessed data.
- QC
- Added normative QC metric based on K-S D statistics calculated for all microarrays in the GEO database.
- Mix-o-matic
- Added confidence intervals by bootstrap for lambda0 estimate.
- Improved design of API.
- Rewritten power analysis code for readability and testability.
- Newer code now has an order of magnitude greater unit test coverage, and produces the same numbers as described for the three datasets in the "Power and sample size estimation in high dimensional biology" Gadbury et al paper.
- http://bugs.ssg.uab.edu/bugzilla/show_bug.cgi?id=153
- Consolidated plotting code for more consistency, this will make it easier to modify and improve in the future.
- More consistent use of visual tweaks and hints for publication-quality plots.
- Additional progress regarding increasing mix-o-matic convergence for more datasets by more sophisticated configuration of the non-linear optimizer and appropriate workarounds when configuration is not an option.
- Deleted residuals
- The plots have been updated for readability for large numbers of arrays per group.
- ANOVA
- Updated missing data handling to properly output a warning message when missing data is detected, and run the F test with a reduced design matrix that does not use the information for the missing elements.
- Updated log window to print warning message ~ WARN - Model is not full rank. Least-squares solutions for the parameters are not unique. Some statistics will be misleading. Redundant columns will be discovered and removed.
- Updated spec and implementation to improve handling for less than full
rank X'X matrix.
- Initial pass at less than full rank X'X matrix handling.
- Input/Ouput component
- Parsers
- Output file handling
Bug Fixes
- Fixed a bug caused by a dataset where, after QQ normalization, the variance for a particular probe is zero.
- Fixed confusing F test interaction term behavior with respect to non-full rank matrices.
- Fixed 'scrunched' interaction term user interface.
Graphical User Interface
- Enhanced the UI to prevent users from selecting FDR or FDRD when they choose to do Chebby checkers statistical test.
- Menubar has been updated with the latest information about HDBStat!
and tooltip functionality added to the menubar.
- Refactored the class that handles printing of messages related to invalid
Java versions.
Web site
- Updated FAQ item to describe what kinds of information to include in a bug report.
- New look & feel
- Better organized for improved user experience.
- Big changes under the hood for readability and clarity, making the web site easier to update and debug in the future.
|
|