Use NMF package to evaluate the optimal number of signatures. This is used along with sig_extract. Users should library(NMF) firstly. If NMF objects are returned, the result can be further visualized by NMF plot methods like NMF::consensusmap() and NMF::basismap().

sig_estimate() shows comprehensive rank survey generated by NMF package, sometimes it is hard to consider all measures. show_sig_number_survey() provides a one or two y-axis visualization method to help users determine the optimal signature number (showing both stability ("cophenetic") and error (RSS) at default). Users can also set custom measures to show.

show_sig_number_survey2() is modified from NMF package to better help users to explore survey of signature number.

sig_estimate(
  nmf_matrix,
  range = 2:5,
  nrun = 10,
  use_random = FALSE,
  method = "brunet",
  seed = 123456,
  cores = 1,
  keep_nmfObj = FALSE,
  save_plots = FALSE,
  plot_basename = file.path(tempdir(), "nmf"),
  what = "all",
  verbose = FALSE
)

show_sig_number_survey(
  object,
  x = "rank",
  left_y = "cophenetic",
  right_y = "rss",
  left_name = left_y,
  right_name = toupper(right_y),
  left_color = "black",
  right_color = "red",
  left_shape = 16,
  right_shape = 18,
  shape_size = 4,
  highlight = NULL
)

show_sig_number_survey2(
  x,
  y = NULL,
  what = c("all", "cophenetic", "rss", "residuals", "dispersion", "evar", "sparseness",
    "sparseness.basis", "sparseness.coef", "silhouette", "silhouette.coef",
    "silhouette.basis", "silhouette.consensus"),
  na.rm = FALSE,
  xlab = "Total signatures",
  ylab = "",
  main = "Signature number survey using NMF package"
)

Arguments

nmf_matrix

a matrix used for NMF decomposition with rows indicate samples and columns indicate components.

range

a numeric vector containing the ranks of factorization to try. Note that duplicates are removed and values are sorted in increasing order. The results are notably returned in this order.

nrun

a numeric giving the number of run to perform for each value in range, nrun set to 30~50 is enough to achieve robust result.

use_random

Should generate random data from input to test measurements. Default is TRUE.

method

specification of the NMF algorithm. Use 'brunet' as default. Available methods for NMF decompositions are 'brunet', 'lee', 'ls-nmf', 'nsNMF', 'offset'.

seed

specification of the starting point or seeding method, which will compute a starting point, usually using data from the target matrix in order to provide a good guess.

cores

number of cpu cores to run NMF.

keep_nmfObj

default is FALSE, if TRUE, keep NMF objects from runs, and the result may be huge.

save_plots

if TRUE, save signature number survey plot to local machine.

plot_basename

when save plots, set custom basename for file path.

what

a character vector whose elements partially match one of the following item, which correspond to the measures computed by summary() on each – multi-run – NMF result: 'all', 'cophenetic', 'rss', 'residuals', 'dispersion', 'evar', 'silhouette' (and more specific *.coef, *.basis, *.consensus), 'sparseness' (and more specific *.coef, *.basis). It specifies which measure must be plotted (what='all' plots all the measures).

verbose

if TRUE, print extra message.

object

a Survey object generated from sig_estimate, or a data.frame contains at least rank columns and columns for one measure.

x

a data.frame or NMF.rank object obtained from sig_estimate().

left_y

column name for left y axis.

right_y

column name for right y axis.

left_name

label name for left y axis.

right_name

label name for right y axis.

left_color

color for left axis.

right_color

color for right axis.

left_shape, right_shape, shape_size

shape setting.

highlight

a integer to highlight a x.

y

for random simulation, a data.frame or NMF.rank object obtained from sig_estimate().

na.rm

single logical that specifies if the rank for which the measures are NA values should be removed from the graph or not (default to FALSE). This is useful when plotting results which include NAs due to error during the estimation process. See argument stop for nmfEstimateRank.

xlab

x-axis label

ylab

y-axis label

main

main title

Value

  • sig_estimate: a list contains information of NMF run and rank survey.

  • show_sig_number_survey: a ggplot object

  • show_sig_number_survey2: a ggplot object

Details

The most common approach is to choose the smallest rank for which cophenetic correlation coefficient starts decreasing (Used by this function). Another approach is to choose the rank for which the plot of the residual sum of squares (RSS) between the input matrix and its estimate shows an inflection point. More custom features please directly use NMF::nmfEstimateRank.

References

Gaujoux, Renaud, and Cathal Seoighe. "A flexible R package for nonnegative matrix factorization." BMC bioinformatics 11.1 (2010): 367.

See also

sig_extract for extracting signatures using NMF package, sig_auto_extract for extracting signatures using automatic relevance determination technique.

sig_estimate for estimating signature number for sig_extract, show_sig_number_survey2 for more visualization method.

Author

Shixiang Wang

Examples

# \donttest{
load(system.file("extdata", "toy_copynumber_tally_M.RData",
  package = "sigminer", mustWork = TRUE
))
library(NMF)
cn_estimate <- sig_estimate(cn_tally_M$nmf_matrix,
  cores = 1, nrun = 5,
  verbose = TRUE
)

p <- show_sig_number_survey2(cn_estimate$survey)
p

# Show two measures
show_sig_number_survey(cn_estimate)
# Show one measure
p1 <- show_sig_number_survey(cn_estimate, right_y = NULL)
p1
p2 <- add_h_arrow(p, x = 4.1, y = 0.953, label = "selected number")
p2

# Show data from a data.frame
p3 <- show_sig_number_survey(cn_estimate$survey)
p3
# Show other measures
head(cn_estimate$survey)
p4 <- show_sig_number_survey(cn_estimate$survey,
  right_y = "dispersion",
  right_name = "dispersion"
)
p4
p5 <- show_sig_number_survey(cn_estimate$survey,
  right_y = "evar",
  right_name = "evar"
)
p5
# }