NEWS
rgsepd 1.27
- library(hash) is going away in 2022, so we need to replace that functionality with native R 4.2 environments. Shouldn't impact users.
rgsepd 1.21
- changing some less than signs to less than or equal to, to ensure we get results.
- planning a refactor to handle other gene ID systems and example datasets. not done
rgsepd 1.15
- new legends in scatterplots.
- custom point colors in PCA plot function. uses sampleMeta$CustomColor and disables the built in legend.
- custom genes in plots. See ExtractProjection() very useful, not yet in vignette - it calculates a projection among your custom gene set.
- Add lfcThreshold to the DESeq routines, so the p-values are adjusted correctly. ( implemented 1.15.1 ) This changes the behavior of most of the system. Expect a different gene set.
- New option to INIT() called renormalize, allowing user to specify preferred matrix normalization schema rather than the default DESeq2::varianceStabilizingTransform (when renormalize=FALSE).
- uses p.adjust(method='BH') on GOSeq's results to shorten the GO term list. use the 'padj' column.
- New option to DESeq , vst with blind=TRUE/FALSE at user discretion.
- added CITATION FILE in 1.15.13
- tweaking vignette in 1.15.15 to debug bioconductor OSX build-server LaTeX versions. And 1.15.16, switching from biblatex to natbib.
rgsepd 1.11.5
BUG FIXES
- updating dependencies caused a crash. Investigating.
GO.db seems to no longer support double-brackets accessor.
Refactored for GO.db 3.6.0's requirement of select()
rgsepd 1.11.2
BUG FIXES
- updating dependencies caused a crash. Investigating.
Biomart seems to no longer support "hsapiens_gene_ensembl" (sometimes/bug)
Intermittent bug fixed in a later biomaRt edition.
- R 3.5.0 support calls for a tweak in GOCatEngine.R
rgsepd 1.7.1
BUG FIXES
- sampleMeta may load as factors, and uses strings later. To prevent possible sample-swapping, need to make all inputs as strings/notfactors. We can force this internally.
rgsepd 1.3.9
BUG FIXES
- Dependency GOSeq 1.23.2 no longer supports gene ID system "refGene", so it has been changed to Entrez Gene IDs under the name "knownGene" instead. Merged gene tables now key off ENTREZ instead of REFSEQ. Now tested and built on R 3.3.0.
rgsepd 1.3.8
FEATURE
- Adjusting the color thresholds for HMG diagram and Gamma scores to be properly scaled by dimensionality. Previous z-scoring method was occasionally making errors. Now a Gamma score of 1 indicates the distance between centroids, which would otherwise naturally scale by the square root of the geneset dimension.
- Updated Vignette
rgsepd 1.3.6
FEATURE
- Simplified final output figure "HMG" and associated tables "HMG1" and "HMG2" calculate the distance to each centroid so the interpretation of results is simpler. Strong overlap to the HMA file.
- Entries in the Vignette for this new 2 table output.
rgsepd 1.3.4
BUG FIXES
- Dependency GenomicRanges no longer exports "assay()"
rgsepd 1.3.3
Rebuilt for R 3.2.2
BUG FIXES
- Biomart.org went down for "ensembl" late 2015. now we ask "www.ensembl.org" directly for their mart. Resolves some warnings, and fixes ID collection.
rgsepd 0.99.15
BUG FIXES
- bugfix in gene ID systems passed to goseq.
rgsepd 0.99.12
NEW FEATURES
- expanding user manuals to clarify how figures and clustering is performed.
rgsepd 0.99.11
BUG FIXES
- bugfix in ProcessAll cardinality message.
- Error catching around PCA.
- Heatmap genelist now respects Annote_Filter file when LIMIT$HARD.
- Cleaned up warnings when no rows for HMA.
- Bugfix in PCA regarding non-negative definite covariance from princomp's eigen: replaced all instances with prcomp().
rgsepd 0.99.10
BUG FIXES
- bugfix in G$LIMIT$baseMean and explanations of naming conventions in the vignette.
rgsepd 0.99.4
SIGNIFICANT USER-VISIBLE CHANGES
- replacing DESeq with DESeq2 Fall2014
rgsepd 0.0.0
NEW FEATURES
- Release package. Fall 2014.