selectFeatures: Select features for MAR/MNAR pattern examination

In DavisLaboratory/msImpute: Imputation of label-free mass spectrometry peptides

Select features for MAR/MNAR pattern examination

Description

Two methods are provided to identify features (peptides or proteins) that can be informative of missing patterns. Method hvp fits a linear model to peptide dropout rate (proportion of samples were peptide is missing) against peptide abundance (average log2-intensity). Method emb is a information theoretic approach to identify missing patterns. It quantifies the heterogeneity (entropy) of missing patterns per biological (experimental group). This is the default method.

Usage

selectFeatures(
  x,
  method = c("ebm", "hvp"),
  group,
  n_features = 500,
  suppress_plot = TRUE
)

Arguments

x	Numeric matrix giving log-intensity where missing values are denoted by NA. Rows are peptides, columns are samples.
method	character. What method should be used to find features? options include method='hvp' and method='ebm'
group	character or factor vector specifying biological (experimental) group e.g. control, treatment, WT, KO
n_features	Numeric, number of features with high dropout rate. 500 by default. Applicable if method="hvp".
suppress_plot	Logical show plot of dropouts vs abundances. Default to TRUE. Applicable if method="hvp".

Details

In general, the presence of group-wise (structured) blocks of missing values, where peptides are missing in one experimental group can indicate MNAR, whereas if such patterns are absent (or missingness is uniform across the samples), peptides are likely MAR. In the presence of MNAR, left-censored MNAR imputation methods should be chosen. Two methods are provided to explore missing patterns: method=hvp identifies top n_features peptides with high average expression that also have high dropout rate, defined as the proportion of samples where peptide is missing. Peptides with high (potentially) biological dropouts are marked in the hvp column in the output dataframe. This method does not use any information about experimental conditions (i.e. group). Another approach to explore and quantify missing patterns is by looking at how homogeneous or heterogeneous missing patterns are in each experimental group. This is done by computing entropy of distribution of observed values. This is the default and recommended method for selectFeatures. Entropy is reported in EBM column of the output. A NaN EBM indicates peptide is missing at least in one experimental group. Features set to TRUE in msImpute_feature column are the features selected by the selected method. Users are encouraged to use the EBM metric to find informative features, hence why the group argument is required.

Value

A data frame with a logical column denoting the selected features

Author(s)

Soroor Hediyeh-zadeh

References

Hediyeh-zadeh, S., Webb, A. I., & Davis, M. J. (2020). MSImpute: Imputation of label-free mass spectrometry peptides by low-rank approximation. bioRxiv.

See Also

msImpute

Examples

data(pxd007959)
group <- pxd007959$samples$group
y <- data.matrix(pxd007959$y)
y <- log2(y)
hdp <- selectFeatures(y, method="ebm", group = group)
# construct matrix M to capture missing entries
M <- ifelse(is.na(y),1,0)
M <- M[hdp$msImpute_feature,]
# plot a heatmap of missingness patterns for the selected peptides
require(ComplexHeatmap)
hm <- Heatmap(M,
column_title = "dropout pattern, columns ordered by dropout similarity",
              name = "Intensity",
              col = c("#8FBC8F", "#FFEFDB"),
              show_row_names = FALSE,
              show_column_names = TRUE,
              cluster_rows = TRUE,
              cluster_columns = TRUE,
              show_column_dend = TRUE,
              show_row_dend = FALSE,
              row_names_gp =  gpar(fontsize = 7),
              column_names_gp = gpar(fontsize = 8),
              heatmap_legend_param = list(#direction = "horizontal",
              heatmap_legend_side = "bottom",
              labels = c("observed","missing"),
              legend_width = unit(6, "cm")),
         )
hm <- draw(hm, heatmap_legend_side = "left")

DavisLaboratory/msImpute documentation built on Jan. 5, 2024, 3:50 a.m.