getJaccardMatrix(), getKappaMatrix(), getMeetMinMatrix() and getSorensenDiceMatrix()
received a notable speedup in their implementation, thanks to the underlying code now
based on the sparse matrix operations implemented in the proxyC package.
Accordingly, the element-wise functions calculateJaccard(), calculateKappa(),
and calculateSorensenDice() were removed from this package.GeneTonic package. Similarly, .map_to_color
handles the mapping of the values to the colors used in the interactive graph views provided
by visNetworkenrichmentWordcloud() has now the option to filter some generic terms, often
encountered in the GO terms collection, as they could risk becoming not too
informative of the underlying processes ( via the remove_generic_terms parameter).
Moreover, any custom term can be removed if specified by the terms_to_remove
parameter.goDistance() now uses the efficient implementation of the distances provided
by the simona package, instead of the fairly cumbersome approach taken by GOSemSimdistanceHeatmap() has now some additional parameter to fine tune the visual
looks of the geneset distance heatmap. The display_similarity parameter can be
set to TRUE to display the similarity values, instead of the distances, Moreover,
the quantile_limits (defaulting to c(0.025, 0.975), the 2.5% and the 97.5%
quantiles) specifies the thresholds that can be used to winsorize the underlying
values - extreme values could lead to a wide usage of most of the color palette,
making patterns harder to discern. This behavior can be deactivated by setting the
quantiles to c(0, 1), which would then include the set of values as-is.path_to_GeDi(), which can be used
to transform the output of various enrichment analysis R packages to be directly
usable in GeDi. The option include packages such as topGO, clusterProfiler,
ReactomePA, enrichR and fgsea.Smaller bug and typo fixes
Fixed the bug that the getGenes() function would set all gene names to all
caps which lead to the inability to download the correct PPI information for
species like mouse. Also renamed the function to prepareGenesetData() to
reflect more accurately its behavior.
Updated the default version used in the getId() and getStringDB() to 12.0,
the current version of the String database.
Fixed the broken zoom feature in the Optional Filtering Step in the Data Input panel. Additionally added a column Description to the table of zoomed gene sets to facilitate interpretation.
Fixed that the clustering will now be reset whenever a new score is calculated.
Updated the checkInclusion() function to drastically reduce runtime.
Fixed the error that the value of alpha would not be properly pass to all the
sub function used by the getpMMMatrix() function.
Replaced all occurrences of PMM with pMM to match the notation of the original publication.
Replaced the kNN clustering algorithm with PAM (partitioning around mendoids) as this seems to be more suitable for enrichment data represented by distance scores.
Renamed the goSimilarity() function to goDistance() to better indicated
that this is a distance rather than a similarity score. Also scaled all
scores to the [0, 1] interval.
Fixed the normalization function in the getKappaDistanceMatrix() function.
Changed the implementation of the Louvain clustering algorithm to use a weighted graph now. The graph is weighted by the distance scores between gene sets.
Added GeneTonicList as a possible input object for GeDi. Now GeDi is directly
compatible with GeneTonic.
col_name_genesets and col_name_genes as parameter to the GeDi() main app to allow users to specify the relevant column names upon executing the command| with || and similar)globalVariables()for loops to avoid unhealthy growing of vectors/matricesSerialParam() to
avoid unexpected behaviors