inst/NEWS.md

In Jfortin1/neuroCombat_Rpackage: Harmonization of multi-site imaging data with ComBat

Changes in Version 1.0.13 - Added neuroCombatData package in Remotes

Changes in Version 1.0.12 - Suggested dependency of newly-created package neuroCombatData - Added vignette using neuroCombatData data

Changes in Version 1.0.11 - Added neuroCombatFromTraining (development version)

Changes in Version 1.0.8 - Fixed a bug related to imaging features for which values are constant within a scanner/batch for EB=FALSE option.

Changes in Version 1.0.7 - Package now depends on BiocParallel. Parellel options available for non-parametric adjustments to speed up computations. - Fixed a bug related to imaging features for which values are constant within a scanner/batch.

Changes in Version 1.0.6 - Changing MIT license to Artistic-2.0 license

Changes in Version 1.0.5 - Started adding training/test functionalities

Changes in Version 1.0.4 - All objects in output have now consistent names (colnames and rownames)

Changes in Version 1.0.3 - In the internal code, we now decouple the standardized mean into two components: intercept (stand.mean) and model mean (mod.mean). The reason is to be able to use stand.mean only when applying scanner correction on a test dataset without making assumption for biological covariates.

Changes in Version 1.0.1: - ComBat now accepts missing values. Note: our way of calculating feature variances (rows) differs from the SVA package. SVA: Variance denominator used when there is no missing values: n SVA: Variance denominator used when there are missing values: m-1 (m=total number of non-missing values) ComBatHarmonization: Variance denominator used when there is no missing values: n ComBatHarmonization: Variance denominator used when there are missing values: m (m=total number of non-missing values) By simulating missing values, we have shown that there is much more better agreement between the datasets.

Jfortin1/neuroCombat_Rpackage documentation built on Nov. 24, 2024, 9:23 a.m.