## psichomics 1.4.0 (22 Oct, 2017)

• Support gene expression data:
• Load, filter, normalise and perform log2-transformation on gene expression data from TCGA
• Perform principal component analysis based on gene expression data, survival analysis by gene expression cutoff and pairwise differential gene expression analysis
• Correlate gene expression of a given gene against PSI values of multiple alternative splicing events
• Filter GTEx junction quantification based on tissues of interest (all tissues are loaded by default)
• Quantify splicing based on a list of genes (splicing events within all genes are quantified by default)
• Parse sample information from TCGA samples using parseTcgaSampleInfo()
• Data grouping:
• Redesigned group creation and selection
• Create groups based on genes and alternative splicing events
• Assign a customisable colour per data group
• Export or import patient and sample identifiers of data groups
• Add new set operations when grouping (such as complement, subtraction and symmetric difference)
• Suggest attributes of interest when creating groups
• Allow to retrieve the universe of patient and sample identifiers by performing the complement group without any group selected
• Statistically analyse group independence (useful to assess the overlap between a PCA cluster and groups derived from clinical and sample attributes, for instance)
• Differential analysis:
• Label points based on top differentially spliced events or genes, selected alternative splicing events and/or selected genes
• Create AS event and gene groups based on filtered or selected AS events and genes in the tables
• Dimensionality reduction techniques:
• Subset data based on groups of AS events and genes before performing dimensionality reduction
• Create data groups based on the partitioning clustering of PCA scores
• Perform independent component analysis (ICA) on alternative splicing quantification and gene expression data
• Survival analysis:
• Add p-value plot to visually infer the significance of survival analyses based on multiple alternative splicing quantification cutoffs
• Gene, transcript and protein information:
• Information retrieval is now only dependent on a user-defined gene, instead of requiring alternative splicing quantification data to be loaded

### Bug fixes and other improvements

• Show progress bar when running in the command-line interface
• Fix inconsistent browser history navigation
• Updated the CLI vignette with information on analysing gene expression data and a quick reference for functions
• Update minimum version required of shiny (1.0.3)
• Avoid replacing selected groups when manipulating new ones
• Differential splicing analysis:
• Fix data not being rendered in the table when zooming in the plot after data transformation was applied
• Return p-value of NA instead of 0 when the value of Fligner-Killeen’s Test for Homogeneity of Variance is infinite
• Discard value transformations that may return invalid data for the values chosen for the X and Y axes
• Fix point that remains highlighted in the plot after deselecting the only selected row of the table
• Improve readability of plot’s tooltip
• Improve survival curves based on the optimal alternative splicing quantification cutoff:
• Include the survival curve previews in 3 new columns within the differential splicing analyses table, instead of below that table; those columns consist of the survival curves, the optimal PSI cutoff and the respective log-rank test’s p-value
• Allow to use survival data when plotting and table sorting
• Include the optimal PSI cutoff and the respective log-rank test’s p-value in exported tables
• Fix link to survival analyses using the previously calculated PSI cutoff
• Principal component analysis:
• When clicking on a alternative splicing event in the loadings plot, the appropriate differential splicing analyses will now be automatically rendered with the respective options, as expected
• Survival analysis:
• Properly set the title of survival curves based on the selected splicing event’s quantification
• Improve readability of Cox PH models
• When performing survival analyses by alternative splicing cutoff, each patient is assigned the PSI value from the respective sample; for patients with more than one sample, the assigned sample is chosen based on the most frequent sample type across all patients (before, the first matched non-normal or non-control samples were used)
• Multiple other bug fixes and visual improvements

## psichomics 1.4.1 (14 Dec, 2017)

• Fix bug that did not allow to select tissues to load GTEx v7 data (graphical interface)
• Fix splicing events not being quantified based on GTEx v7 junction reads
• Gene expression normalisation:
• Fix misleading gene expression (non-)normalisation by converting reads to counts per million (CPM) using edgeR::cpm() after normalisation using edgeR::calcNormFactor()
• Alternative splicing quantification:
• Updated support to properly parse new notation of alternative splicing annotation from Bioconductor (backwards compatible with older notation)
• Raise error when no splicing events after quantification
• Fix warning following the quantification of splicing events or its loading (incorrect parsing of gene information from splicing events)
• Dimensionality reduction:
• Use the number (instead of the percentage) of tolerated missing values per sample as the argument to impute data from the remaining samples for those values before performing dimensionality reduction; by default, missing values are tolerated for 10 samples
• Update file description and README
• Minor bug fixes and improvements

## psichomics 1.4.2 (19 Dec, 2017)

• Fix error when trying to load alternative splicing annotation (given updated hg19 and hg38 annotation that is now available for use with psichomics)

## psichomics 1.4.3 (12 Jan, 2018)

• Alternative splicing quantification:
• Improve speed and memory usage when dealing with larger datasets
• Improve quantification of alternative first and last exons: quantify alternative first and last exons based on all exon-exon junction reads that support each of the alternative exons
• Print progress bar in R console
• Principal component analysis:
• Change tolerated missing values per event to 5% by default
• Show/save table with the contribution of events (for alternative splicing quantification) or genes (for gene expression) to the selected principal components
• Allow to plot top 100 variables that most contribute to the selected principal components (faster rendering of and interaction with loading plots)
• Differential analysis:
• Allow to input a list of groups for the “group” argument of the functions diffAnalyses() and plotDistribution() (command-line interface)
• Allow to input a non-numeric vector or a row of a matrix/data frame in the “data” argument of the function plotDistribution() (command-line interface)
• Correlation between gene expression and alternative splicing:
• Perform correlation between gene expression of multiple genes and quantification of multiple alternative splicing events

### Bug fixes

• Fix unnamed events when only one event for a event type is returned
• Minor copy-editing and overall improvements

## psichomics 1.4.4 (12 Feb, 2018)

• Update CITATION file to show citation to article in bioRxiv: https://www.biorxiv.org/content/early/2018/02/07/261180
• Update vignettes to include a case study based on the aforementioned article
• Gene expression pipeline:
• Perform limma-trend by default
• Alternative splicing quantification:
• Quantification of alternative first and last exons: following more thorough testing, the new exon-centred method was considered to be less relevant to exploratory analysis (specially when compared with other types of events); as such, both methods are now available for quantification
• Dimensionality reduction:
• Change tolerated missing values per event to 5% by default for both PCA and ICA (in both visual and command-line interfaces)
• PCA: In the table that shows the events that most contribute to the selected principal components, show the rank
• Groups:
• Automatically set sample type groups (i.e. normal, primary solid tumour, metastatic, etc.) for TCGA samples (visual interface only)
• Differential analysis:
• Use numeric fields instead of sliders to precisely filter data
• Fix table for differential expression not being filtered based on highlighted genes
• Filter splicing events or genes to use when performing exploratory differential analyses
• Survival analysis:
• Allow to stratify patients based on optimal gene expression cutoff
• Select samples to be used for survival analysis

### Bug fixes

• Fix problems related with DT versions >= 0.3:
• Groups displayed as having the same attributes as last created group
• Table not being updated in differential analyses according to event filtering based on the volcano plot

## psichomics 1.4.5 (4 Apr, 2018)

• Mention psichomics manuscript throughout psichomics
• Copy-edit graphical user interface and respective tutorial
• Fix warnings and errors in Bioconductor