createPositives: createPositives

In mcalgaro93/benchdamic: Benchmark of differential abundance methods on microbiome data

createPositives

Description

Inspect the list of p-values or/and log fold changes from the output of the differential abundance detection methods and count the True Positives (TP) and the False Positives (FP).

Usage

createPositives(
  object,
  priorKnowledge,
  enrichmentCol,
  namesCol = NULL,
  slot = "pValMat",
  colName = "adjP",
  type = "pvalue",
  direction = NULL,
  threshold_pvalue = 1,
  threshold_logfc = 0,
  top = NULL,
  alternative = "greater",
  verbose = FALSE,
  TP,
  FP
)

Arguments

object	Output of differential abundance detection methods. pValMat, statInfo matrices, and method's name must be present (See vignette for detailed information).
priorKnowledge	data.frame (with feature names as row.names) containing feature level metadata.
enrichmentCol	name of the column containing information for enrichment analysis.
namesCol	name of the column containing new names for features (default namesCol = NULL).
slot	A character vector with 1 or number-of-methods-times repeats of the slot names where to extract values for each method (default slot = "pValMat").
colName	A character vector with 1 or number-of-methods-times repeats of the column name of the slot where to extract values for each method (default colName = "rawP").
type	A character vector with 1 or number-of-methods-times repeats of the value type of the column selected where to extract values for each method. Two values are possible: "pvalue" or "logfc" (default type = "pvalue").
direction	A character vector with 1 or number-of-methods-times repeats of the statInfo's column name containing information about the signs of differential abundance (usually log fold changes) for each method (default direction = NULL).
threshold_pvalue	A single or a numeric vector of thresholds for p-values. If present, features with p-values lower than threshold_pvalue are considered differentially abundant. Set threshold_pvalue = 1 to not filter by p-values.
threshold_logfc	A single or a numeric vector of thresholds for log fold changes. If present, features with log fold change absolute values higher than threshold_logfc are considered differentially abundant. Set threshold_logfc = 0 to not filter by log fold change values.
top	If not null, the top number of features, ordered by p-values or log fold change values, are considered as differentially abundant (default top = NULL).
alternative	indicates the alternative hypothesis and must be one of "two.sided", "greater" or "less". You can specify just the initial letter. Only used in the 2 \times 2 case.
verbose	Boolean to display the kind of extracted values (default verbose = FALSE).
TP	A list of length-2 vectors. The entries in the vector are the direction ("UP Abundant", "DOWN Abundant", or "non-DA") in the first position, and the level of the enrichment variable (enrichmentCol) which is expected in that direction, in the second position.
FP	A list of length-2 vectors. The entries in the vector are the direction ("UP Abundant", "DOWN Abundant", or "non-DA") in the first position, and the level of the enrichment variable (enrichmentCol) which is not expected in that direction, in the second position.

Value

a data.frame object which contains the number of TPs and FPs features for each method and for each threshold of the top argument.

See Also

getPositives, plotPositives.

Examples

data("ps_plaque_16S")
data("microbial_metabolism")

# Extract genera from the phyloseq tax_table slot
genera <- phyloseq::tax_table(ps_plaque_16S)[, "GENUS"]
# Genera as rownames of microbial_metabolism data.frame
rownames(microbial_metabolism) <- microbial_metabolism$Genus
# Match OTUs to their metabolism
priorInfo <- data.frame(genera,
    "Type" =  microbial_metabolism[genera, "Type"])
# Unmatched genera becomes "Unknown"
unknown_metabolism <- is.na(priorInfo$Type)
priorInfo[unknown_metabolism, "Type"] <- "Unknown"
priorInfo$Type <- factor(priorInfo$Type)
# Add a more informative names column
priorInfo[, "newNames"] <- paste0(rownames(priorInfo), priorInfo[, "GENUS"])

# Add some normalization/scaling factors to the phyloseq object
my_norm <- setNormalizations(fun = c("norm_edgeR", "norm_CSS"),
    method = c("TMM", "CSS"))
ps_plaque_16S <- runNormalizations(normalization_list = my_norm,
    object = ps_plaque_16S)
# Initialize some limma based methods
my_limma <- set_limma(design = ~ 1 + RSID + HMP_BODY_SUBSITE, 
    coef = "HMP_BODY_SUBSITESupragingival Plaque",
    norm = c("TMM", "CSS"))

# Make sure the subject ID variable is a factor
phyloseq::sample_data(ps_plaque_16S)[, "RSID"] <- as.factor(
    phyloseq::sample_data(ps_plaque_16S)[["RSID"]])
    
# Perform DA analysis
Plaque_16S_DA <- runDA(method_list = my_limma, object = ps_plaque_16S)

# Count TPs and FPs, from the top 1 to the top 20 features.
# As direction is supplied, features are ordered by "logFC" absolute values.
positives <- createPositives(object = Plaque_16S_DA,
    priorKnowledge = priorInfo, enrichmentCol = "Type", 
    namesCol = "newNames", slot = "pValMat", colName = "rawP", 
    type = "pvalue", direction = "logFC", threshold_pvalue = 1, 
    threshold_logfc = 0, top = 1:20, alternative = "greater", 
    verbose = FALSE,
    TP = list(c("DOWN Abundant", "Anaerobic"), c("UP Abundant", "Aerobic")),
    FP = list(c("DOWN Abundant", "Aerobic"), c("UP Abundant", "Anaerobic")))

# Plot the TP-FP differences for each threshold
plotPositives(positives = positives)

mcalgaro93/benchdamic documentation built on March 10, 2024, 10:40 p.m.