Version 4.0

• There is now an option to use k-means clustering for non-temporal data. Unlike the other two methods, k-means does not use the threshold parameter. Instead, the number of clusters is chosen based on the dimension of the input matrix. The number of clusters is varied, and the silhouette index is used to choose an optimal number. If you wish to change the cluster number for k-means, you may alter the kmeans_clustering.R function.
• There is now an option to not normalize edge weights. Previously, edge weights were always normalized to the [0,1] range prior to trimming. This results in more edges in the network, as edge weights with negative values are transformed prior to trimming. Not performing normalization will result in a smaller network, as edge weights with negative values will be automatically trimmed.
• Parallelization is now implemented in the RS.Get.Weight.Matrix.R function - the RS.Get.Weight.Matrix_cluster.R function is no longer necessary. The number of cores for parallelization can be specified in the RShiny App. Leaving the number of cores at the default value (1) will disable parallelization.
• All prior tutorials and test datasets can still be used, but the UI will now appear slightly different due to the newly implemented parameters.