Interpreting Cluster Means/Centers for High Dimensionality Clustering Using Kmeans Clusting

See also questions close to this topic

• Rscript: problems with inner_join

  I am trying to run my R-script from command line, but it is return error. Where are no one problems, when I run similar code from R or Rstudio. Before, I faced a problem with merge (Rcript Error in if (nx >= 2^31 || ny >= 2^31) stop("long vectors are not supported")) and I have replaced it to inner_join from dplyr

  Error in UseMethod("tbl_vars") :
    no applicable method for 'tbl_vars' applied to an object of class "function"
  Calls: inner_join ... tbl_vars -> new_sel_vars -> structure -> tbl_vars_dispatch
  

  The problematic part of R-script:

  clonotypes_tables = function(name, cell, mode){
    sub = subset(metadata, metadata$donor == as.character(name))
    sub =  subset(sub, sub$cell_type == as.character(cell))
    if (nrow(sub) > 1){
      sub = sub[order(sub$time_point), ]
    
      if (file.exists(paste(getwd(), sub$file_name[1], sep="/")) & file.exists(paste(getwd(), sub$file_name[2], sep="/"))){
        point1 = read.table(sub$file_name[1], header = T)
        #cat("check1")
        point2 = read.table(sub$file_name[2], header = T)
        same_src(point1, point2)
      
        if (nrow(point1) >= 1000 & nrow(point2) >= 1000){
          if (mode == "CDR3_V"){
            
            common.clonotype = inner_join(point1[,c(1,2,4,5)], point2[,c(1,2,4,5)], by = c("cdr3aa", "v"), copy = T)
            common.clonotype$clon = paste(common.clonotype$cdr3aa, common.clonotype$v, sep = "~")
          }
          else{
            common.clonotype = inner_join(point1[,c(1,2,4)], point2[,c(1,2,4)], by = c("cdr3aa"), copy = T)
            common.clonotype$clon = common.clonotype$cdr3aa
          }
          common.clonotype = common.clonotype[,c("clon", "freq.x", "freq.y")]
          colnames(common.clonotype) = c("Clonotypes", "0.5", "1")
          dim(common.clonotype)
          common.clonotype = common.clonotype[order(common.clonotype[2], decreasing = T), ]
          common.clonotype
        }
        #return(common.clonotype)
      }
    else{
      print(paste(name, cell, "hasn't two time points", sep = " ")) 
     }
    }
  }
  

  Also, I have tried inner_join(point1[,c(1,2,4)], point2[,c(1,2,4)], by = c("cdr3aa")), but where was problem with src without copy = T
• Mclust (GMM model) clustering in R: how can I add label for each point in the clusters?

  I would like to perform a model-based clustering based on parameterized finite Gaussian mixture models. I used Mclust function by mclust package in R. I have a dataframe like this (which I reported here for the first 10 rows):

  df=data.frame(X1=c(-0.9749422,
                 -0.4062771,
                 1.0974262,
                 -2.297264,
                 -1.6022011,
                 2.144254,
                 2.2012879,
                 2.810878,
                 0.6728063,-0.7042836), X2= c(0.1740782,
                                              -1.4475989,
                                              0.8575626,
                                              1.8466605,
                                              -2.0279622,
                                              2.9748541,
                                              0.8820755,
                                              3.0898032,
                                              1.3757168,
                                              -1.9729475),
            label=as.character(c("abundant",
                                 "accelerating",
                                 "anyone",
                                 "approach",
                                 "association",
                                 "ban",
                                 "blog",
                                 "commission",
                                 "complete",
                                 "congratulations")),stringsAsFactors = FALSE)
  

  i.e.

  >   df      
      X1          X2          label
  1  -0.9749422    0.1740782  abundant
  2  -0.4062771   -1.4475989  accelerating
  3   1.0974262    0.8575626  anyone
  4  -2.297264     1.8466605  approach
  5  -1.6022011   -2.0279622  association
  6   2.144254     2.9748541  ban
  7   2.2012879    0.8820755  blog
  8   2.810878     3.0898032  commission
  9   0.6728063    1.3757168  complete
  1  -0.7042836   -1.9729475  congratulations
  

  I would like to plot the clusters obtained by Mclust function in X1 X2 plane, indicating the corresponding label of each point in the cluster, then I started from (3 clusters e.g.):

  mod2 <- Mclust(plot_df[,1:2], G = 3)
  plot(mod2,"classification")
  

  but obviously I obtained no labels. How can I add label for each point in the clusters?

  Thank you!
• Make columns on the basis of one column

  I have a data set in R as following:-

  a <- data.frame(name=c("AFG", "AFG", "AFG","AFG", "GER", "GER", "GER", "GER", "GFR", "GFR", 
                        "GFR", "GFR"),
              Typ = c("One", "Two", "Three", "Four", "One", "Two", "Three", "Four", "One", 
                     "Two", "Three", "Four"),
              Yr1 = c(10, 11, 12, 14, 15, 17, 18, 19, 88, 1, 39, 1),
              Yr2 = c(1:12), Yr3 = c(8:19))
  

  I want to change this data so that the columns are based on the values in Typ column. That is I get the following data.frame.

  b <- data.frame(name = c("AFG", "AFG", "AFG","GER", "GER", "GER","GFR", "GFR", "GFR"),
              Yr = c("Yr1", "Yr2", "Yr3", "Yr1", "Yr2", "Yr3", "Yr1", "Yr2", "Yr3"),
              One = c(10, 1, 8, 15, 5, 12, 88, 9, 16),
              Two = c(11, 2, 9, 17, 6, 13, 1, 10, 17),
              Three = c(12, 3, 10, 18, 7, 14, 39, 11, 18),
              Four = c(14, 4, 11, 19, 8, 15, 1, 12, 19))
  

  Thanks in advance