Produces boxplots to visualise the distribution of the selected genes. For each dataset, you can choose which clinical variable to group the samples onWhen choosing Cambridge or Stockholm, you will have the option to display the expression in the five different subtypes identified by Ross-Adams et al (2015). These subtypes were shown to have significantly different outcomesIf multiple microarray probes are found for the gene, the probe with the highest inter-quartile range (IQR) will be pickedAn ANOVA analysis will also be performed to assess whether there are different expression levels in the groups you have chosenThe boxplot can be exported as a pdf or png image. An R script can be downloaded, allowing you to repeat the analysis or tweak as you wish
You can perform Recursive Partitioning on a selected gene in a dataset with survival information (Cambridge, Stockholm and MSKCC). This analysis will determine if there are sub-groups of samples with significantly different expression levelIf samples in the dataset can be allocated into different groups based on the expression of the gene, a Kaplan-Meir plot will be displayed. Otherwise, the median expression level of the gene will be used to assign samples to high and low expression groups
You can plot one gene against another in a specified dataset. Points on the plot are coloured according to sample group. The correlation is computed and displayed.
The uploaded gene list can be used to generate a heatmap from the chosen dataset. Control is given over the distance metric and clustering method. Samples can be partitioned into different groups based on the clustering, and the composition of each group can be interrogated
For datasets with Copy number information (Cambridge, Stockholm and MSKCC), the frequency of alterations in different clinical covariates is displayed. A heatmap can also be generatedWe are very grateful to Emilie Lalonde from University of Toronto for supplying the data for these plots
Spinning Wait Icons by Andrew Davidson http://andrewdavidson.com/articles/spinning-wait-icons/
Here we show the results of an ANOVA (analysis of variance) analysis to assess whether there are changes in expression level between the defined groups
A recursive partitioning (RP) analysis  is performed to determine if the samples can be split into groups based on the expression data from your chosen gene(s). An RP p-value < 0.05 indicates a significant split. The p-value from RP and cut-off corresponding to a split are shown in the table belowIf no cut-off can be found with RP, the samples will be divided according to median expression level in the plots belowA histogram of expression level will be shown with a line to indicate the median expression level or RP cut-offThe grouping of samples found by RP, or using median expression level, is used to construct a Kaplan-Meier plotThe Kaplan-Meier plot is a useful way of summarising survival data. There is one curve for each group. Each curve starts at 100% probability of survival. The probability of freedom from biochemical recurrence is shown on the y axis and the time (in years) is shown on the x axis. The curve drops each time there is an 'event'. A cross is shown on each curve where a 'censoring'' event takes place. This is where someone drops out of the study for a reason not related to the study, e.g. the study ends before an event has occurred. These subjects are no longer included in any calculations. The lower the survival curve the worse prognosis the patients in that group have.
Construcing a heatmap from the gene list you uploaded in the Analysis Parameters tab. If you haven't uploaded a gene list, an example gene list of three genes will be usedSample ClusteringSelect the number of clusters, k, from the slider
The Proportion of amplifications and deletions will be shown for your chosen gene(s).