auxfunc.R 15 KB
Newer Older
dmattek's avatar
dmattek committed
1
## Custom plotting
dmattek's avatar
dmattek committed
2
require(ggplot2)
dmattek's avatar
Mod:  
dmattek committed
3 4 5
require(RColorBrewer)
require(gplots) # for heatmap.2
require(grid) # for modifying grob
dmattek's avatar
dmattek committed
6

dmattek's avatar
dmattek committed
7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
rhg_cols <- c(
  "#771C19",
  "#AA3929",
  "#E25033",
  "#F27314",
  "#F8A31B",
  "#E2C59F",
  "#B6C5CC",
  "#8E9CA3",
  "#556670",
  "#000000"
)

md_cols <- c(
  "#FFFFFF",
  "#F8A31B",
  "#F27314",
  "#E25033",
  "#AA3929",
  "#FFFFCC",
  "#C2E699",
  "#78C679",
  "#238443"
)

dmattek's avatar
dmattek committed
32 33 34 35 36 37 38 39 40 41 42 43 44
s.cl.linkage = c("ward.D",
                 "ward.D2",
                 "single",
                 "complete",
                 "average",
                 "mcquitty",
                 "centroid")

s.cl.spar.linkage = c("average",
                      "complete", 
                      "single",
                      "centroid")

dmattek's avatar
Added:  
dmattek committed
45
s.cl.diss = c("euclidean", "maximum", "manhattan", "canberra", "binary", "minkowski", "DTW")
dmattek's avatar
dmattek committed
46 47
s.cl.spar.diss = c("squared.distance","absolute.value")

48
# list of palettes for the heatmap
dmattek's avatar
dmattek committed
49 50 51 52 53 54 55 56 57 58
l.col.pal = list(
  "White-Orange-Red" = 'OrRd',
  "Yellow-Orange-Red" = 'YlOrRd',
  "Reds" = "Reds",
  "Oranges" = "Oranges",
  "Greens" = "Greens",
  "Blues" = "Blues",
  "Spectral" = 'Spectral'
)

59 60 61 62 63 64 65 66 67 68
# list of palettes for the dendrogram
l.col.pal.dend = list(
  "Rainbow" = 'rainbow_hcl',
  "Sequential" = 'sequential_hcl',
  "Heat" = 'heat_hcl',
  "Terrain" = 'terrain_hcl',
  "Diverge HCL" = 'diverge_hcl',
  "Diverge HSV" = 'diverge_hsv'
)

dmattek's avatar
Added:  
dmattek committed
69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97
# Creates a popup with help text
# From: https://gist.github.com/jcheng5/5913297
helpPopup <- function(title, content,
                      placement=c('right', 'top', 'left', 'bottom'),
                      trigger=c('click', 'hover', 'focus', 'manual')) {
  tagList(
    singleton(
      tags$head(
        tags$script("$(function() { $(\"[data-toggle='popover']\").popover(); })")
      )
    ),
    tags$a(
      href = "#", class = "btn btn-mini", `data-toggle` = "popover",
      title = title, `data-content` = content, `data-animation` = TRUE,
      `data-placement` = match.arg(placement, several.ok=TRUE)[1],
      `data-trigger` = match.arg(trigger, several.ok=TRUE)[1],
      #tags$i(class="icon-question-sign")
      # changed based on http://stackoverflow.com/questions/30436013/info-bubble-text-in-a-shiny-interface
      icon("question")
    )
  )
}

help.text = c(
  'Accepts CSV file with a column of cell IDs for removal. 
                   IDs should correspond to those used for plotting. 
  Say, the main data file contains columns Metadata_Site and TrackLabel. 
  These two columns should be then selected in UI to form a unique cell ID, e.g. 001_0001 where former part corresponds to Metadata_Site and the latter to TrackLabel.',
  'Plotting and data processing requires a unique cell ID across entire dataset. A typical dataset from CellProfiler assigns unique cell ID (TrackLabel) within each field of view (Metadata_Site).
98 99 100
                   Therefore, a unique ID is created by concatenating these two columns. If the dataset already contains a unique ID, UNcheck this box and select a single column only.',
  'This option allows to interpolate NAs or missing data. Some rows in the input file might be missing because a particular time point might not had been acquired. 
  This option, interpolates such missing points as well as points with NAs in the measurement column. When this option is checked, the interval of time column must be provided!'
dmattek's avatar
Added:  
dmattek committed
101 102 103 104
)


#####
dmattek's avatar
dmattek committed
105
## Functions for clustering 
dmattek's avatar
Added:  
dmattek committed
106

dmattek's avatar
dmattek committed
107 108 109 110 111 112 113 114 115

# Return a dt with cell IDs and corresponding cluster assignments depending on dendrogram cut (in.k)
# This one works wth dist & hclust pair
# For sparse hierarchical clustering use getDataClSpar
# Arguments:
# in.dend  - dendrogram; usually output from as.dendrogram(hclust(distance_matrix))
# in.k - level at which dendrogram should be cut

getDataCl = function(in.dend, in.k) {
dmattek's avatar
Added:  
dmattek committed
116 117
  cat(file = stderr(), 'getDataCl \n')
  
dmattek's avatar
dmattek committed
118 119 120 121 122 123 124 125
  loc.m = dendextend::cutree(in.dend, in.k, order_clusters_as_data = TRUE)
  #print(loc.m)
  
  # The result of cutree containes named vector with names being cell id's
  # THIS WON'T WORK with sparse hierarchical clustering because there, the dendrogram doesn't have original id's
  loc.dt.cl = data.table(id = names(loc.m),
                         cl = loc.m)
  
126 127
  #cat('===============\ndataCl:\n')
  #print(loc.dt.cl)
dmattek's avatar
dmattek committed
128
  return(loc.dt.cl)
dmattek's avatar
Added:  
dmattek committed
129 130
}

dmattek's avatar
dmattek committed
131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149

# Return a dt with cell IDs and corresponding cluster assignments depending on dendrogram cut (in.k)
# This one works with sparse hierarchical clustering!
# Arguments:
# in.dend  - dendrogram; usually output from as.dendrogram(hclust(distance_matrix))
# in.k - level at which dendrogram should be cut
# in.id - vector of cell id's

getDataClSpar = function(in.dend, in.k, in.id) {
  cat(file = stderr(), 'getDataClSpar \n')
  
  loc.m = dendextend::cutree(in.dend, in.k, order_clusters_as_data = TRUE)
  #print(loc.m)
  
  # The result of cutree containes named vector with names being cell id's
  # THIS WON'T WORK with sparse hierarchical clustering because there, the dendrogram doesn't have original id's
  loc.dt.cl = data.table(id = in.id,
                         cl = loc.m)
  
150 151
  #cat('===============\ndataCl:\n')
  #print(loc.dt.cl)
dmattek's avatar
dmattek committed
152 153 154 155 156
  return(loc.dt.cl)
}



dmattek's avatar
Added:  
dmattek committed
157 158 159 160 161 162 163 164 165 166 167 168 169 170 171
# prepares a table with cluster numbers in 1st column and colour assignments in 2nd column
# the number of rows is determined by dendrogram cut
getClCol <- function(in.dend, in.k) {
  
  loc.col_labels <- get_leaves_branches_col(in.dend)
  loc.col_labels <- loc.col_labels[order(order.dendrogram(in.dend))]
  
  return(unique(
    data.table(cl.no = dendextend::cutree(in.dend, k = in.k, order_clusters_as_data = TRUE),
               cl.col = loc.col_labels)))
}


#####
## Common plotting functions
dmattek's avatar
dmattek committed
172

dmattek's avatar
Mod:  
dmattek committed
173 174 175 176 177 178 179 180 181 182 183 184
myGgplotTraj = function(dt.arg, # data table
                        x.arg,  # string with column name for x-axis
                        y.arg, # string with column name for y-axis
                        group.arg, # string with column name for grouping time series (typicaly cell ID)
                        facet.arg, # string with column name for facetting
                        facet.ncol.arg = 2, # default number of facet columns
                        facet.color.arg = NULL, # vector with list of colours for adding colours to facet names (currently a horizontal line on top of the facet is drawn)
                        line.col.arg = NULL, # string with column name for colouring time series (typically when individual time series are selected in UI)
                        xlab.arg = NULL, # string with x-axis label
                        ylab.arg = NULL, # string with y-axis label
                        plotlab.arg = NULL, # string with plot label
                        dt.stim.arg = NULL, # plotting additional dataset; typically to indicate stimulations (not fully implemented yet, not tested!)
dmattek's avatar
dmattek committed
185
                        tfreq.arg = 1,
dmattek's avatar
dmattek committed
186
                        ylim.arg = NULL,
dmattek's avatar
dmattek committed
187
                        stim.bar.height.arg = 0.1,
dmattek's avatar
Added:  
dmattek committed
188
                        stim.bar.width.arg = 0.5,
dmattek's avatar
Mod:  
dmattek committed
189
                        aux.label1 = NULL, # 1st point label; used for interactive plotting; displayed in the tooltip; typically used to display values of column holding x & y coordinates
dmattek's avatar
Added:  
dmattek committed
190
                        aux.label2 = NULL,
191
                        aux.label3 = NULL,
dmattek's avatar
Added:  
dmattek committed
192 193 194 195 196
                        stat.arg = c('', 'mean', 'CI', 'SE')) {
  
  # match arguments for stat plotting
  loc.stat = match.arg(stat.arg, several.ok = TRUE)

dmattek's avatar
Added:  
dmattek committed
197 198
  
  # aux.label12 are required for plotting XY positions in the tooltip of the interactive (plotly) graph
dmattek's avatar
dmattek committed
199 200
  p.tmp = ggplot(dt.arg,
                 aes_string(x = x.arg,
dmattek's avatar
dmattek committed
201
                            y = y.arg,
dmattek's avatar
Added:  
dmattek committed
202
                            group = group.arg,
203 204 205 206 207
                            label = group.arg))
  #,
  #                          label  = aux.label1,
  #                          label2 = aux.label2,
  #                          label3 = aux.label3))
dmattek's avatar
dmattek committed
208
  
dmattek's avatar
dmattek committed
209 210 211 212 213 214 215 216 217 218 219 220 221
  if (is.null(line.col.arg)) {
    p.tmp = p.tmp +
      geom_line(alpha = 0.25, 
                              size = 0.25)
  }
  else {
    p.tmp = p.tmp + 
      geom_line(aes_string(colour = line.col.arg), 
                              alpha = 0.5, 
                              size = 0.5) +
      scale_color_manual(name = '', 
                         values =c("FALSE" = rhg_cols[7], "TRUE" = rhg_cols[3], "SELECTED" = 'green', "NOT SEL" = rhg_cols[7]))
  }
dmattek's avatar
Mod:  
dmattek committed
222 223 224 225 226 227 228 229 230 231

  # this is temporary solution for adding colour according to cluster number
  # use only when plotting traj from clustering!
  # a horizontal line is added at the top of data
  if (!is.null(facet.color.arg)) {

    loc.y.max = max(dt.arg[, c(y.arg), with = FALSE])
    loc.dt.cl = data.table(xx = 1:length(facet.color.arg), yy = loc.y.max)
    setnames(loc.dt.cl, 'xx', facet.arg)
    
dmattek's avatar
Fixed:  
dmattek committed
232 233
    # adjust facet.color.arg to plot
    
dmattek's avatar
Mod:  
dmattek committed
234 235 236 237 238
    p.tmp = p.tmp +
      geom_hline(data = loc.dt.cl, colour = facet.color.arg, yintercept = loc.y.max, size = 4) +
      scale_colour_manual(values = facet.color.arg,
                          name = '')
  }
dmattek's avatar
dmattek committed
239
  
dmattek's avatar
Added:  
dmattek committed
240 241
  if ('mean' %in% loc.stat)
    p.tmp = p.tmp + 
dmattek's avatar
dmattek committed
242 243 244
    stat_summary(
      aes_string(y = y.arg, group = 1),
      fun.y = mean,
dmattek's avatar
Added:  
dmattek committed
245
      colour = 'red',
dmattek's avatar
dmattek committed
246 247 248 249
      linetype = 'solid',
      size = 1,
      geom = "line",
      group = 1
dmattek's avatar
Added:  
dmattek committed
250 251 252 253 254 255 256 257
    )

  if ('CI' %in% loc.stat)
    p.tmp = p.tmp + 
    stat_summary(
      aes_string(y = y.arg, group = 1),
      fun.data = mean_cl_normal,
      colour = 'red',
dmattek's avatar
Mod:  
dmattek committed
258
      alpha = 0.25,
dmattek's avatar
Added:  
dmattek committed
259 260 261 262 263 264 265 266 267 268
      geom = "ribbon",
      group = 1
    )
  
  if ('SE' %in% loc.stat)
    p.tmp = p.tmp + 
    stat_summary(
      aes_string(y = y.arg, group = 1),
      fun.data = mean_se,
      colour = 'red',
dmattek's avatar
Mod:  
dmattek committed
269
      alpha = 0.25,
dmattek's avatar
Added:  
dmattek committed
270 271 272 273 274 275 276
      geom = "ribbon",
      group = 1
    )
  
  
  
  p.tmp = p.tmp + 
dmattek's avatar
dmattek committed
277 278 279 280 281 282 283 284 285 286 287 288 289 290 291
    facet_wrap(as.formula(paste("~", facet.arg)),
               ncol = facet.ncol.arg,
               scales = "free_x")
  
  if(!is.null(dt.stim.arg)) {
    p.tmp = p.tmp + geom_segment(data = dt.stim.arg,
                                 aes(x = Stimulation_time - tfreq.arg,
                                     xend = Stimulation_time - tfreq.arg,
                                     y = ylim.arg[1],
                                     yend = ylim.arg[1] + abs(ylim.arg[2] - ylim.arg[1]) * stim.bar.height.arg),
                                 colour = rhg_cols[[3]],
                                 size = stim.bar.width.arg,
                                 group = 1) 
  }
  
dmattek's avatar
dmattek committed
292 293 294
  if (!is.null(ylim.arg)) 
    p.tmp = p.tmp + coord_cartesian(ylim = ylim.arg)
  
dmattek's avatar
dmattek committed
295 296 297 298
  p.tmp = p.tmp + 
    xlab(paste0(xlab.arg, "\n")) +
    ylab(paste0("\n", ylab.arg)) +
    ggtitle(plotlab.arg) +
299
    ggplotTheme() + 
300
    theme(legend.position = "top")
dmattek's avatar
dmattek committed
301
  
dmattek's avatar
Mod:  
dmattek committed
302
  return(p.tmp)
dmattek's avatar
dmattek committed
303 304 305
}


306 307 308 309 310 311 312
# Fast DTW computation
fastDTW <-function (x)
{
  return(dtw(x, window.type = 'sakoechiba', distance.only = T)$normalizedDistance)
}


dmattek's avatar
dmattek committed
313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344
# Plots a scatter plot with marginal histograms
# Points are connected by a line (grouping by cellID)
#
# Assumes an input of data.table with
# x, y - columns with x and y coordinates
# id - a unique point identifier (here corresponds to cellID)
# mid - a (0,1) column by which points are coloured (here corresponds to whether cells are within bounds)

myGgplotScat = function(dt.arg,
                        band.arg = NULL,
                        facet.arg = NULL,
                        facet.ncol.arg = 2,
                        xlab.arg = NULL,
                        ylab.arg = NULL,
                        plotlab.arg = NULL,
                        alpha.arg = 1,
                        group.col.arg = NULL) {
  p.tmp = ggplot(dt.arg, aes(x = x, y = y))
  
  if (is.null(group.col.arg)) {
    p.tmp = p.tmp +
      geom_point(alpha = alpha.arg, aes(group = id))
  } else {
    p.tmp = p.tmp +
      geom_point(aes(colour = as.factor(get(group.col.arg)), group = id), alpha = alpha.arg) +
      geom_path(aes(colour = as.factor(get(group.col.arg)), group = id), alpha = alpha.arg) +
      scale_color_manual(name = group.col.arg, values =c("FALSE" = rhg_cols[7], "TRUE" = rhg_cols[3], "SELECTED" = 'green'))
  }
  
  if (is.null(band.arg))
    p.tmp = p.tmp +
      stat_smooth(
dmattek's avatar
dmattek committed
345 346 347
        # method = function(formula, data, weights = weight)
        #   rlm(formula, data, weights = weight, method = 'MM'),
        method = "lm",
dmattek's avatar
dmattek committed
348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389
        fullrange = FALSE,
        level = 0.95,
        colour = 'blue'
      )
  else {
    p.tmp = p.tmp +
      geom_abline(slope = band.arg$a, intercept = band.arg$b) +
      geom_abline(
        slope = band.arg$a,
        intercept =  band.arg$b + abs(band.arg$b)*band.arg$width,
        linetype = 'dashed'
      ) +
      geom_abline(
        slope = band.arg$a,
        intercept = band.arg$b - abs(band.arg$b)*band.arg$width,
        linetype = 'dashed'
      )
  }
  
  if (!is.null(facet.arg)) {
    p.tmp = p.tmp +
      facet_wrap(as.formula(paste("~", facet.arg)),
                 ncol = facet.ncol.arg)
    
  }
  
  
  if (!is.null(xlab.arg))
    p.tmp = p.tmp +
      xlab(paste0(xlab.arg, "\n"))
  
  if (!is.null(ylab.arg))
    p.tmp = p.tmp +
      ylab(paste0("\n", ylab.arg))
  
  if (!is.null(plotlab.arg))
    p.tmp = p.tmp +
      ggtitle(paste0(plotlab.arg, "\n"))
  
  
  
  p.tmp = p.tmp +
390
    ggplotTheme() +
391 392
    theme(legend.position = "none")

dmattek's avatar
dmattek committed
393 394 395 396 397 398
  # Marginal distributions don;t work with plotly...
  # if (is.null(facet.arg))
  #   ggExtra::ggMarginal(p.scat, type = "histogram",  bins = 100)
  # else
  return(p.tmp)
}
dmattek's avatar
dmattek committed
399

400

dmattek's avatar
Mod:  
dmattek committed
401 402 403 404 405 406 407 408 409 410 411 412 413
myPlotHeatmap <- function(data.arg,
                          dend.arg,
                          palette.arg,
                          palette.rev.arg = TRUE,
                          dend.show.arg = TRUE,
                          key.show.arg = TRUE,
                          margin.x.arg = 5,
                          margin.y.arg = 20,
                          nacol.arg = 0.5,
                          colCol.arg = NULL,
                          labCol.arg = NULL,
                          font.row.arg = 1,
                          font.col.arg = 1,
414
                          breaks.arg = NULL,
dmattek's avatar
Mod:  
dmattek committed
415 416
                          title.arg = 'Clustering') {
  
417 418
  loc.n.colbreaks = 99
  
dmattek's avatar
Mod:  
dmattek committed
419 420
  if (palette.rev.arg)
    my_palette <-
421
    rev(colorRampPalette(brewer.pal(9, palette.arg))(n = loc.n.colbreaks))
dmattek's avatar
Mod:  
dmattek committed
422 423
  else
    my_palette <-
424
    colorRampPalette(brewer.pal(9, palette.arg))(n = loc.n.colbreaks)
dmattek's avatar
Mod:  
dmattek committed
425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459
  
  
  col_labels <- get_leaves_branches_col(dend.arg)
  col_labels <- col_labels[order(order.dendrogram(dend.arg))]
  
  if (dend.show.arg) {
    assign("var.tmp.1", dend.arg)
    var.tmp.2 = "row"
  } else {
    assign("var.tmp.1", FALSE)
    var.tmp.2 = "none"
  }
  
  loc.p = heatmap.2(
    data.arg,
    Colv = "NA",
    Rowv = var.tmp.1,
    srtCol = 90,
    dendrogram = var.tmp.2,
    trace = "none",
    key = key.show.arg,
    margins = c(margin.x.arg, margin.y.arg),
    col = my_palette,
    na.col = grey(nacol.arg),
    denscol = "black",
    density.info = "density",
    RowSideColors = col_labels,
    colRow = col_labels,
    colCol = colCol.arg,
    labCol = labCol.arg,
    #      sepcolor = grey(input$inPlotHierGridColor),
    #      colsep = 1:ncol(loc.dm),
    #      rowsep = 1:nrow(loc.dm),
    cexRow = font.row.arg,
    cexCol = font.col.arg,
dmattek's avatar
dmattek committed
460 461
    main = title.arg,
    symbreaks = FALSE,
462 463
    symkey = FALSE,
    breaks = if (is.null(breaks.arg)) NULL else seq(breaks.arg[1], breaks.arg[2], length.out = loc.n.colbreaks+1)
dmattek's avatar
Mod:  
dmattek committed
464 465 466 467
  )
  
  return(loc.p)
}