R/fct_calculate_tsne.R

In JohannesSchroth/CytoDA: Cytometry Data Analysis

#' Function to calculate tSNE
#' 

calculate_tsne <- function(input, output, session, data, vars, dims, cluster_df) {
  
  withProgress(message = 'Calculating tSNE', value = 0, {
    
    incProgress(0.2)
    
    set.seed(123)
    
    tsne <- Rtsne::Rtsne(X = data[,vars], dims = which_dim(dims), verbose = T,
                         perplexity = input$perplexity_tsne, 
                         theta = input$theta_tsne, 
                         max_iter = input$max_iter_tsne)
    incProgress(0.8)
    colnames(tsne$Y) <- c('tSNE1', 'tSNE2', 'tSNE3')[1:which_dim(dims)]
    
    tsne <- as.data.frame(cbind(data, tsne$Y))
    
    if(is.null(cluster_df) != TRUE) {
      tsne <- cbind(tsne, cluster_df)
    }
    
    updateSelectInput(session = session, inputId = 'colour_col2', label = 'Colour by:', choices = colnames(tsne), selected = 'Rphenograph_Clusters')
    
    output$tsne_plot <- renderPlotly({
      
      plot_ly(data = tsne, x = tsne$tSNE1, y = tsne$tSNE2, z = if (dims == '3D') tsne$tSNE3 else NULL) %>%
        add_markers(color = ~tsne[,input$colour_col2]) %>%
        layout(dragmode = 'lasso') %>%
        layout(font = list(size = 25)) %>%
        hide_colorbar() 
    })
    incProgress(1)
  })
  output$download_tsne_data <- downloadHandler(
    
    filename = function() {
      paste('tsne_data_', Sys.Date(), ".csv", sep = "")
    },
    content = function(file) {
      write.csv(tsne, file, row.names = FALSE)
    }
    
  )
  return(tsne)
}