A visualisation tool for the creation and analysis of graphs
Add facility to select favourite transforms, which are displayed at the top of the transform list
Add mechanism to save sets of transforms and visualisations for later recall
The correlation plot columns can now be sorted by data value
The correlation plot is now vertically scrollable when it's too large for its container - i.e. when there are many column annotations
In the correlation loader, input data can now be clipped against a constant, or by Winsorization
Correlation plot columns can now be sorted by hierarchical clustering, using the SLINK method with a Euclidean distance measure
Add a Filtering option to correlation parameters, more specifically enabling the ability to filter edges using the k-Nearest Neighbours strategy; this is particularly useful with datasets that are extremely correlated, where traditional thresholding cannot adequately differentiate between rows
Replace the Initial Transforms section of the correlation loader with Template, allowing for a more flexible and customisable means of appending additional transforms and visualisations
Add ability to save correlation plot images on a per attribute value basis, e.g. saving one image per cluster
Downloadable files can now be opened directly, either by using the Open Url option in the File menu, or by dragging and dropping hyperlinks onto the main window
A Graphia specific URL scheme is now registered so that the application may now be opened by clicking special hyperlinks; these can be created using the Link Generator
Tabular lists of node pairs can now be loaded from pairwise CSV, TSV, SSV and XLSX files
Loading Cytoscape eXchange files is now possible, as used on NDex
Graphviz .dot files may now be loaded
When loading from an adjacency matrix, it is now possible to choose to discard edges based on edge weight, and to skip double edges where the matrix is symmetric
Visualisations that use deleted source attributes can now have their source attributes changed, in contrast to the prior behaviour where a broken visualisation would need to be deleted and recreated
For numerical visualisations, the mapping of input value to output value can now be fine tuned
An Apply button is now present on the gradient, palette and numeric mapping visualisation dialogs, for immediate preview of any changed settings
Attributes can now be imported, accessible via Tools → Import Attributes From Table…
User defined attributes can now be removed if desired, accessible via Tools → Remove Attributes…
Attributes can now be cloned, accessible via Tools → Clone Attribute…
User defined attributes can now be edited, accessible via Tools → Edit Attribute…
There is now search functionality in all the places where attributes are chosen; this is most useful when selecting a Data Value attribute in the case where there are many
The node attribute table now has a context sensitive menu that provides options relevant to the clicked attribute
The contents of a column from the node attribute table can now be copied to the clipboard, accessible by the context menu
Layout settings are now accessible via a toolbar button
Node and edge sizes are now saved in the .graphia file and are adjustable via Layout Settings
Layout settings can now be saved as named presets, for quick access to frequently used settings
Correlating tables of discrete values is now possible, using the Jaccard Index or Simple Matching Coefficient algorithms
For continuous correlation, the Euclidean Distance, Cosine Similarity and Biweight Midcorrelation (Bicor) algorithms are now available
Column annotation colours are now assigned by palette rather than by hashing the string value
Numerical column annotations are now displayed using a gradient
In the data plot, values can now be scaled by an attribute value
The epsilon used when log scaling data is now computed to use an appropriately sized value rather than a fixed constant
When loading correlation data with duplicate attribute column names, these are now considered separately rather than overwriting each other
The data plot can now be panned by dragging and zoomed using the mouse wheel
In the plot, data can now be combined into IQR boxplots based on one or more annotations, enabled by selecting Group By Annotation in the Plot menu
Furthermore, when Group By Annotation is enabled there is now another option to colour the IQR boxes with a user selected annotation
When displaying IQR plots, outliers can now be disabled by toggling the Show Outliers menu option
Various buttons have been added to the tool bar to control the correlation plot
Data can now be log scaled when rendering IQR boxplots
Hyperlinks in table cells are now clickable and some basic HTML 3.2 can be displayed
A log is now made of the steps taken in the graph's creation, accessible via Tools → Show Provenance Log…
There is now an option to stay in component mode when finding, by default off
The Latest Changes option was added to the Help menu, which shows any new changes present in the running version