- Changed
theme_coriell()
to drop panel borders and removed angle from x-axis text. - Changed
plot_volcano()
andplot_md()
defaults to use unaliased points, smaller text sizes for labels, and moved labels further to the sides. - Added some significance testing to the
pairwise_intersections()
. - Added a new function for plotting pairwise correlations between columns in a
matrix,
plot_cor_pairs()
- Added an argument for
panther_go()
to include the reference gene list when performing over representation testing. There may still be bugs here. - Updated site. Removed some vignettes with older workflows. TODO: add new vignettes with current best practices for RNAseq.
- Added new arguments for
plot_volcano()
andplot_md()
to allow setting axis limits prior to determining annotation placement. These functions now also usetheme_coriell()
by default.
- Updated
meta_de()
function to operate strictly on SummarizedExperiment objects. This function is significantly faster than the previous version. - Added a helper function,
dfs2se()
to convert a list of data.frames to a SummarizedExperiment object for use bymeta_de()
- Added a helper function to perform jackknife resampling on the columns of a SummarizedExperiment,
jackknifeSE()
.
- Removed meta-analysis functions,
meta_vote()
,meta_pcombine()
andplot_metavolcano()
, in favor of the newermeta_de()
function which provides an interface tometapod
for combining p-values in a more robust way. - Removed a redundant scaling step in
quickmap()
when calculating breaks and made some changes to the way thefix_extreme
argument behaves. - Updated RNA-seq article with more analysis steps and helper functions
- Updated
panther_go()
to usehttr2
- Added
UMAP()
andplot_umap()
functions. TheUMAP()
functions accepts PCA objects fromPCAtools
,prcomp
, or a distance matrix or raw data matrix and exposes theumap.defaults
as function arguments. - The
plot_umap()
function provides a simple plotting method for the data.frame produced by theUMAP()
function.
- Internal changes to the
quickmap()
function. Avoidpheatmap
scaling in favor of vectorized scaling. Speed up removeVar calculations withRfast::rowVars()
ormatrixStats::rowVars()
if available. Speed up clustering and distance matrix calculations by performing distance matrix calculations withrdist::rdist()
and clustering withfastcluster::hclust()
if available. Round values in fix_extreme to better maintain original scale limits.
- Moved most packages to "Suggests" instead of "Imports" to reflect that this package is a collection of helpers. This reduces dependencies upon install.
- Inclusion of and update for
plot_boxplot()
,plot_density()
, andplot_parallel()
. These functions are now generics that work with matrix, data.frame, andSummarizedExperiment
classes
- Potential breaking changes to
plot_volcano()
andplot_md()
. Forplot_volcano()
set the default value for the labels to NULL and removed the removed the labels altogether forplot_md()
- Added new plotting functions for expression matrices:
plot_boxplot()
,plot_density()
, andplot_parallel()
- Added a new ggplot2 theme,
theme_coriell()
- Set defaults on
plot_volcano()
andplot_md()
to have consistent colors - Added new argument to
quickmap()
that removes low variance features before plotting.
- Added function for performing pairwise fisher tests relative to a reference.
pairwise_fisher_test()
- Added arguments to
quickmap()
to enable fixing the colors at the extreme ends of the data. - Added lab_size arguments to
plot_volcano()
andplot_md()
- Added
rarefy()
function. Replacessubsample_counts()
- Removed some old functions.
- Added new function
read_bismark()
that reads in a list of Bismark coverage files and optionally filters by coverage and variance.
- Updated
plot_md()
,plot_volcano()
, andsummarize_dge()
to removedplyr()
dependency. The changes to these functions are breaking. Arguments for column names must now be quoted. plot_volcano()
andplot_md()
now support additional arguments for modifying the point size, shape, and color. See function documentation.
- Added a function for calling outlier values in columns of a numeric matrix by the IQR method
- Removed
process_quant_file()
function. Switched to usingtximport()
in all pipelines.
- Added meta-analysis functions that rip off
MetaVolcanoR
but are much faster.- The
meta_vote()
function implements a vote-counting strategy for determining common differentially expressed genes - The
meta_pcombine()
function combines p-values and logFCs across studies. - The
plot_metavolcano()
function provides a plotting function specific to themeta_vote()
results.
- The
- Eliminated export of
magittr
pipe. Nowcoriell
doesn't export the pipe.
edger_to_df()
returns a data.frame instead of a tibble- New functions for generating random color palettes:
distinct_rgb_palette()
andrandom_rgb_palette()
- Added function for defining threshold based on ranked data:
rank_threshold()
. Inspired by unimodal thresholding algorithm from image analysis. panther_go()
now returns adata.table
of the raw, unlisted data returned from the request. The original version pivoted the table wider using columns for the GO term and the description of the GO term. This result gave inaccurate results when using a different pathway in the function call.- General move towards reducing the number of dependencies in functions by either
removing outside packages and switching to base R or moving to
data.table()
. - Added utility functions for transforming numeric values.
- Updated internals for
permutation_correlation_test()
that applies permuted vector in a vectorized fashion over the entire matrix. - Removed filtering capability from
edger_to_df()
to allow for anyEdgeR
results objects to be used as input.