Changelog • coriell

coriell 0.15.0

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() and plot_md() to allow setting axis limits prior to determining annotation placement. These functions now also use theme_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 by meta_de()
Added a helper function to perform jackknife resampling on the columns of a SummarizedExperiment, jackknifeSE().

Removed meta-analysis functions, meta_vote(), meta_pcombine() and plot_metavolcano(), in favor of the newer meta_de() function which provides an interface to metapod 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 the fix_extreme argument behaves.
Updated RNA-seq article with more analysis steps and helper functions
Updated panther_go() to use httr2

Added UMAP() and plot_umap() functions. The UMAP() functions accepts PCA objects from PCAtools, prcomp, or a distance matrix or raw data matrix and exposes the umap.defaults as function arguments.
The plot_umap() function provides a simple plotting method for the data.frame produced by the UMAP() function.

Internal changes to the quickmap() function. Avoid pheatmap scaling in favor of vectorized scaling. Speed up removeVar calculations with Rfast::rowVars() or matrixStats::rowVars() if available. Speed up clustering and distance matrix calculations by performing distance matrix calculations with rdist::rdist() and clustering with fastcluster::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(), and plot_parallel(). These functions are now generics that work with matrix, data.frame, and SummarizedExperiment classes

Potential breaking changes to plot_volcano() and plot_md(). For plot_volcano() set the default value for the labels to NULL and removed the removed the labels altogether for plot_md()
Added new plotting functions for expression matrices: plot_boxplot(), plot_density(), and plot_parallel()

Added a new ggplot2 theme, theme_coriell()
Set defaults on plot_volcano() and plot_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() and plot_md()
Added rarefy() function. Replaces subsample_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(), and summarize_dge() to remove dplyr() dependency. The changes to these functions are breaking. Arguments for column names must now be quoted.
plot_volcano() and plot_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 using tximport() 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 the meta_vote() results.
Eliminated export of magittr pipe. Now coriell 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() and random_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 a data.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 any EdgeR results objects to be used as input.