CHANGED: major rewrite of the occupancy FDR logic for gene expression status. Briefly:
Model regression fits have been improved with weighted natural spline fits for Tier 2 regressions (relationship is not log-linear)
Threshold scores for null model fitting are now determined from input data
Diagnostic plots for Tier 2 regressions have been added
Empirical p-values now aggregated on a condition level by default, using either Stouffer’s (default) or Fisher’s method
Occupancy p-values are now calculated by default during data loading, and merged during differential_binding(), before BH adjustment is applied to the aggregated p-values per condition. See the relevant man pages for more details.
CHANGED: internal logic for handling condition names and matches
FIXED: Venn diagram universe logic now correctly determines the expressed gene universe when handling RNA Polymerase datasets
damidBind 0.99.10
NEW: plot diagnostics for sample loading (PCA, clustered correlation heatmap)
NEW: plot diagnostics for limma functions (eBayes moderation / SA plots)
NEW: limma functions now use trend and robust by default to better fit heteroscedastic DamID data
NEW: new function extract_unique_sample_ids generates simplifed, unique sample names from complex filenames for display (in diagnostic plots, IGV/Shiny browser)
CHANGED: dense point labelling function now uses exact dbscan kNN functions, not approximate HNSW functions
CHANGED: differential threshold filtering can now be set to non-zero values
CHANGED: default replicate filtering for differential analyses is set to the minimum number of the two condition replicates
damidBind 0.99.9
FIXED: igvShiny code now correctly handles the new internal GRanges binding profile data objects
damidBind 0.99.8
Unicode removed from documentation to allow latex generation :/
damidBind 0.99.6
NEW: Differential analyses now optionally screen out loci with negative signal in a specified number of replicates per sample (default is 2). These loci are removed before differential analysis.
NEW: Geneset universe calculations now take into account the low-signal screening; universe is determined only on the loci passing filter.
NEW: Loci labels are optionally sampled to allow labels even in highly-dense plot regions. Sampling algorithm determines a KNN graph from points, then uses this to select only a subset of total labelled points. A default parameter is provided, but all algorithm parameters are fully customisable.
NEW: Test conditions specified via the cond parameter can now use regexes, when regex = TRUE is set.
CHANGED: Test conditions specified via cond now use a simple named vector, merging cond and cond_names.
CHANGED: Improved and fixed underlying logic and tests to handle the cond parameter change
CHANGED: Substantially improved legend guide handling parameters and logic for volcano plots
FIXED: a number of minor bugfixes, code refactoring and tidying.
damidBind 0.99.4
NEW: Added in FDR modelling and calculation routines for RNA Polymerase occupancy
NEW: plot_volcano and plot_venn functions can now filter for FDR when plotting
NEW: expressed() accessor method for obtaining all genes passing an FDR threshold for a specific condition
CHANGED: binding profile data is now stored as a GRanges object
CHANGED: analyse_go_terms now returns the clusterProfiler results object for downstream use
Begun incorporating gseGO functionality into analyse_go_terms
Improved documentation of DamIDResults class and accessors
Added more details to the package vignette
Many small bug fixes
damidBind 0.99.2
NEW: Added button in IGV viewer to save SVG (useful when viewing window is horizontally compressed and button in the IGV interface is not displayed)
NEW: New legend customisation options for volcano plots (including a legend internal to the plot frame, now default)
NEW: More GO term plot customisation options
FIXED: All volcano plot point labels are now rendered as a single geom layer, preventing ugly overlaps