Package 'lncRna'

Aggregate Coding Potential Tool Results

Description

Reads output files from various coding potential prediction tools and aggregates their results into a structured list. Non-coding predictions are encoded as 1, and coding predictions as 0.

Usage

aggregateCodPot(
  CPC2_outfile = NULL,
  PLEK_outfile = NULL,
  FEELnc_outfile = NULL,
  CPAT_outfile = NULL,
  CPAT_cutoff = 0.364,
  CNCI_outfile = NULL,
  LncFinder_outfile = NULL,
  lncRNA_Mdeep_outfile = NULL
)

Arguments

CPC2_outfile	Path to the CPC2 output file.
PLEK_outfile	Path to the PLEK output file.
FEELnc_outfile	Path to the FEELnc output file.
CPAT_outfile	Path to the CPAT output file.
CPAT_cutoff	Cutoff value for CPAT probability (default: 0.364).
CNCI_outfile	Path to the CNCI output file.
LncFinder_outfile	Path to the LncFinder output file.
lncRNA_Mdeep_outfile	Path to the lncRNA_Mdeep output file.

Value

A list containing seqIDs (a character vector of all unique sequence IDs found across all provided files) and tools (a list where each element is a binary vector of predictions for a given tool).

Examples

# --- 1. Create temporary output files for a reproducible example ---
cpc2File <- tempfile()
plekFile <- tempfile()

# CPC2 format: ID ... classification
write.table(
  data.frame(V1="ID1", V8="noncoding"),
  cpc2File, col.names=FALSE, row.names=FALSE, sep="\t"
)
write.table(
  data.frame(V1="ID2", V8="coding"),
  cpc2File, col.names=FALSE, row.names=FALSE, sep="\t", append=TRUE
)

# PLEK format: Classification Score >ID
write.table(
  data.frame(V1="Non-coding", V2="0.9", V3=">ID2"),
  plekFile, col.names=FALSE, row.names=FALSE, sep=" "
)

# --- 2. Run the function with the temporary files ---
codpotResults <- aggregateCodPot(CPC2_outfile = cpc2File, PLEK_outfile = plekFile)

print("Sequence IDs:")
print(codpotResults$seqIDs)
print("Tool predictions:")
print(codpotResults$tools)

# --- 3. Clean up the temporary files ---
unlink(c(cpc2File, plekFile))

---

Annotate Genomic Interactions with Functional Enrichment Results

Description

Merges a table of genomic interactions (e.g., lncRNA-mRNA) with functional enrichment results (e.g., from g:Profiler), annotating each interaction with relevant functional terms.

Usage

annotateInteractions(
  gostResult,
  interactionTable,
  type,
  lncRnaCol = "lncRNAId",
  targetCol = "targetRNAId"
)

Arguments

gostResult	A data.frame from a g:Profiler analysis (gost function), containing at least "term_name" and "intersection" columns.
interactionTable	A data.frame of interactions. Must contain columns for lncRNA and target identifiers.
type	A character string specifying the interaction type to be added to the output (e.g., "cis", "trans", "LncTar").
lncRnaCol	An optional character string specifying the column name in interactionTable for lncRNA IDs. Defaults to lncRNAId.
targetCol	An optional character string specifying the column name in interactionTable for target RNA IDs. Defaults to targetRNAId.

Value

A data.frame where enrichment terms are merged with the interactions they are associated with.

Examples

# --- 1. Create mock input data ---
# a) Mock g:Profiler result
mockGostResult <- data.frame(
  term_name = c("response to stress", "cell activation"),
  intersection = c("TARGET_G1,TARGET_G2", "TARGET_G3"),
  source = c("GO:BP", "GO:BP"),
  stringsAsFactors = FALSE
)

# b) Mock interaction table with standard column names
mockInteractionTable1 <- data.frame(
  lncRNAId = c("LNC_G1", "LNC_G2", "LNC_G3"),
  targetRNAId = c("TARGET_G1", "TARGET_G2", "TARGET_G3")
)

# --- 2. Run with standard column names ---
annotated_trans <- annotateInteractions(
  gostResult = mockGostResult,
  interactionTable = mockInteractionTable1,
  type = "trans"
)
print(annotated_trans)

# --- 3. Run with custom column names ---
# a) Mock interaction table with custom column names
mockInteractionTable2 <- data.frame(
  Query = "LNC_G1",
  Target = "TARGET_G1"
)

annotated_lnctar <- annotateInteractions(
  gostResult = mockGostResult,
  interactionTable = mockInteractionTable2,
  type = "LncTar",
  lncRnaCol = "Query",
  targetCol = "Target"
)
print(annotated_lnctar)

---

Evaluate Performance of Coding Potential Tools

Description

Computes a comprehensive set of confusion matrix statistics (e.g., accuracy, sensitivity, specificity) for individual coding potential prediction tools. The function operates on summary data prepared by summarizeTestSet.

Usage

bestTool(summaryList, tools = NULL, digits = 4)

Arguments

summaryList	A list generated by summarizeTestSet(), containing ⁠$tools⁠ (a list of 0/1 prediction vectors) and ⁠$isNC⁠ (a numeric vector of true labels: 1=nc, 0=cds).
tools	An optional character vector specifying the names of the tools (from names(summaryList$tools)) to analyze. If NULL, the behavior depends on the session type: in an interactive session, the user will be prompted to select tools; in a non-interactive session, all available tools will be analyzed by default.
digits	The number of decimal places to round the final statistics to (default: 4).

Value

A data frame where rows correspond to performance metrics and columns correspond to the selected tools. Returns NULL if the input is invalid or no statistics can be calculated.

Examples

# --- 1. Create a mock object mimicking the output of prepareEvaluationSets ---
# This object contains filtered data ready for evaluation.
set.seed(123)
n_seq <- 100
exampleEvaluationSummary <- list(
  tools = list(
    CPC2 = sample(c(0, 1), n_seq, replace = TRUE),
    CPAT = sample(c(0, 1), n_seq, replace = TRUE),
    PLEK = sample(c(0, 1), n_seq, replace = TRUE)
  ),
  isNC = sample(c(0, 1), n_seq, replace = TRUE) # True labels
)

# --- 2. Run bestTool ---
# Example 1: Analyze a specific subset of tools
performance_stats <- bestTool(
  summaryList = exampleEvaluationSummary,
  tools = c("CPC2", "CPAT")
)
print(performance_stats)

# Example 2: Analyze all available tools (non-interactively)
all_tools_stats <- bestTool(summaryList = exampleEvaluationSummary)
print(all_tools_stats)

# --- 3. Interactive example (do not run in scripts) ---
if (interactive()) {
  # The following would prompt you to select tools from the console:
  # interactive_stats <- bestTool(summaryList = exampleSummaryList)
}

---

Evaluate Performance of Agreement Thresholds

Description

Computes confusion matrix statistics for each "at least n" agreement threshold generated by evaluateToolsThresholds.

Usage

bestToolAtleast(agreementSummary, digits = 4)

Arguments

agreementSummary	A list generated by evaluateToolsThresholds(), containing ⁠$isNC⁠ (true labels) and ⁠$atLeastN⁠ (a list of 0/1 prediction vectors named atl1, atl2, etc.).
digits	The number of decimal places to round the final statistics to (default: 4).

Value

A data frame where rows are performance metrics and columns correspond to the "at least n" (atl1, atl2, ...) thresholds. Returns NULL if the input is invalid or no statistics can be calculated.

Examples

# --- 1. Create a mock object mimicking the output of prepareEvaluationSets ---
# This object serves as input for evaluating thresholds.
set.seed(789)
n_seq <- 100
# First, create the summary object (as if from prepareEvaluationSets)
evaluationSummary <- list(
  seqIDs = paste0("S", 1:n_seq),
  isNC = sample(c(0, 1), n_seq, replace = TRUE),
  type = sample(c("nc", "cds"), n_seq, replace = TRUE),
  tools = list(
    ToolA = sample(c(0, 1), n_seq, replace = TRUE),
    ToolB = sample(c(0, 1), n_seq, replace = TRUE),
    ToolC = sample(c(0, 1), n_seq, replace = TRUE)
  )
)

# --- 2. Create the agreement summary using evaluateToolsThresholds ---
# This is the object that bestToolAtleast actually takes as input
agreementSummary <- evaluateToolsThresholds(summaryList = evaluationSummary)

# --- 3. Run bestToolAtleast on the result ---
if (!is.null(agreementSummary)) {
  performance_thresholds <- bestToolAtleast(
    agreementSummary = agreementSummary
  )
  print(performance_thresholds)
}

---

Evaluate Performance of Tool Combinations

Description

Computes confusion matrix statistics for predictions generated by various tool combinations, using the output from evaluateToolCombinations.

Usage

bestToolCombination(combinationSummaryList, combinations = NULL, digits = 4)

Arguments

combinationSummaryList	A list generated by evaluateToolCombinations(), containing ⁠$isNC⁠ (true labels) and ⁠$toolCombinations⁠ (a list of prediction vectors for each combination).
combinations	An optional character vector specifying the names of the tool combinations (e.g., "ToolA+ToolB") to analyze. If NULL, behavior depends on the session: interactive mode prompts for selection, while non-interactive mode analyzes all available combinations.
digits	The number of decimal places to round final statistics to (default: 4).

Value

A data frame where rows are performance metrics and columns correspond to the analyzed tool combinations. Returns NULL on error or if no statistics can be calculated.

Examples

# --- 1. Create a mock object mimicking evaluateToolCombinations output ---
set.seed(202)
n_seq <- 80
exampleCombinationSummary <- list(
  isNC = sample(c(0, 1), n_seq, replace = TRUE),
  toolCombinations = list(
    `ToolA+ToolB` = sample(c(0, 1), n_seq, replace = TRUE, prob = c(0.7, 0.3)),
    `ToolA+ToolC` = sample(c(0, 1), n_seq, replace = TRUE, prob = c(0.6, 0.4)),
    `ToolB+ToolC` = sample(c(0, 1), n_seq, replace = TRUE, prob = c(0.5, 0.5))
  )
  # Other elements like seqIDs, type would also be present
)

# --- 2. Run the function ---
# Example 1: Analyze specific combinations
perf_specific <- bestToolCombination(
  combinationSummaryList = exampleCombinationSummary,
  combinations = c("ToolA+ToolB", "ToolB+ToolC")
)
print("Performance for specific combinations:")
print(perf_specific)

# Example 2: Analyze all combinations (non-interactively)
perf_all <- bestToolCombination(
  combinationSummaryList = exampleCombinationSummary
)
print("Performance for all combinations:")
print(perf_all)

---

Calculate and Filter Confusion Matrices for Tool Combinations

Description

This function calculates confusion matrices for various tool combinations based on predictions from evaluateToolCombinations. It can also filter these matrices based on pre-calculated performance metrics and a given threshold.

Usage

calculateCM(
  combinationSummaryList,
  metricsData = NULL,
  printMetricThresholds = FALSE,
  threshold = 0.8,
  returnOnlyHigh = FALSE,
  metricToExtract = "Accuracy"
)

Arguments

combinationSummaryList	A list generated by evaluateToolCombinations(), containing ⁠$toolCombinations⁠ (a list of 0/1 prediction vectors) and ⁠$isNC⁠ (the true class labels).
metricsData	An optional data frame where rows are metric names and columns are tool combination names (typically the output of bestToolCombination()). If provided, metricToExtract is looked up here for threshold-based filtering and printing. Defaults to NULL.
printMetricThresholds	Logical. If TRUE, prints the value of metricToExtract for each combination, indicating if it meets the threshold. Defaults to FALSE.
threshold	A numeric value (0 to 1) used to evaluate metricToExtract. Defaults to 0.8.
returnOnlyHigh	Logical. If TRUE, returns confusion matrices only for combinations meeting the threshold. Defaults to FALSE.
metricToExtract	The performance metric to use for filtering and printing (e.g., "Accuracy", "Sensitivity"). Defaults to "Accuracy".

Value

A named list where each element represents a confusion matrix for a tool combination. Each element is itself a list containing:

table	A 2x2 numeric matrix representing the confusion table.
positive	A character string indicating the positive class level ("1").
metrics	A named numeric vector of all calculated performance metrics.

The returned list is filtered if returnOnlyHigh = TRUE.

Examples

# --- 1. Create mock data mimicking the outputs of previous functions ---
set.seed(202)
n_seq <- 100
# a) Mock output from evaluateToolCombinations()
mockCombinationSummary <- list(
  isNC = sample(c(0, 1), n_seq, replace = TRUE),
  toolCombinations = list(
    `ToolA+ToolB` = sample(c(0,1), n_seq, replace=TRUE, prob=c(0.2,0.8)),
    `ToolA+ToolC` = sample(c(0,1), n_seq, replace=TRUE, prob=c(0.6,0.4)),
    `ToolB+ToolC` = sample(c(0,1), n_seq, replace=TRUE, prob=c(0.5,0.5))
  )
)
# b) Mock output from bestToolCombination()
mockMetricsData <- bestToolCombination(
  combinationSummaryList = mockCombinationSummary
)

# --- 2. Run calculateCM in different modes ---
# Example 1: Calculate all CMs without filtering
all_cms <- calculateCM(combinationSummaryList = mockCombinationSummary)
print(names(all_cms))
# Inspect the structure of a single element
str(all_cms[[1]])

# Example 2: Print and filter based on a Sensitivity threshold >= 0.5
filtered_cms <- calculateCM(
  combinationSummaryList = mockCombinationSummary,
  metricsData = mockMetricsData,
  printMetricThresholds = TRUE,
  returnOnlyHigh = TRUE,
  metricToExtract = "Sensitivity",
  threshold = 0.5
)
print("Filtered CMs (Sensitivity >= 0.5):")
print(names(filtered_cms))

---

Split a Set of Sequence Names into Training and Test Sets

Description

This function partitions a vector of sequence names into training and testing subsets based on a specified percentage. It uses random sampling, so for reproducible splits, set a seed with set.seed() before calling it. This single function replaces the previous, separate test.train.cds and test.train.nc functions.

Usage

createTrainTestSets(sequences, percentTrain = 0.6, prefix = "set")

Arguments

sequences	A character vector of sequence names, or an object with a names attribute (like the list returned by seqinr::read.fasta).
percentTrain	A numeric value between 0 and 1 indicating the proportion of sequences to allocate to the training set (default: 0.6).
prefix	A character string used as a prefix for the names of the elements in the returned list (default: "set"). For example, prefix = "cds" will result in list elements named "cds.train" and "cds.test".

Value

A list containing two character vectors. The names of the list elements are constructed using the prefix argument (e.g., cds.train and cds.test).

Examples

# --- Example 1: Splitting CDS sequences (replaces test.train.cds) ---
all_cds_names <- paste0("cds_seq_", 1:200)
set.seed(123) # for a reproducible split
cds_split <- createTrainTestSets(
  sequences = all_cds_names,
  percentTrain = 0.7,
  prefix = "cds"
)
names(cds_split)
length(cds_split$cds.train)
length(cds_split$cds.test)

# --- Example 2: Splitting non-coding sequences (replaces test.train.nc) ---
# Input can also be a list with names
nc_fasta_like <- as.list(paste0("nc_seq_", 1:100))
names(nc_fasta_like) <- paste0("nc_seq_", 1:100)
set.seed(456)
nc_split <- createTrainTestSets(
  sequences = nc_fasta_like,
  percentTrain = 0.5,
  prefix = "nc"
)
names(nc_split)
length(nc_split$nc.train)
length(nc_split$nc.test)

---

Evaluate Performance of Tool Combinations

Description

Identifies sequences predicted as non-coding by specific combinations (intersections) of tools. It generates prediction vectors for all combinations of 2 or more selected tools.

Usage

evaluateToolCombinations(summaryList, tools = NULL)

Arguments

summaryList	A list generated by prepareEvaluationSets(), containing ⁠$seqIDs⁠, ⁠$tools⁠ (a list of 0/1 prediction vectors), ⁠$isNC⁠, and ⁠$type⁠.
tools	An optional character vector specifying which tools to use for creating combinations. If NULL, the behavior depends on the session: interactive mode will prompt for selection, while non-interactive mode will use all available tools.

Value

A list containing:

seqIDs	Original sequence identifiers from the input.
isNC	Original true labels (1=nc, 0=cds) from the input.
type	Original type annotation ('nc', 'cds') from the input.
selectedToolsPredictions	A list of prediction vectors for the tools selected for this analysis.
toolCombinations	A list where each element is named after a tool combination (e.g., "ToolA+ToolB") and contains a binary vector (0/1) indicating if a sequence was predicted as non-coding by ALL tools in that combination.

Examples

# --- 1. Create a mock object mimicking the output of prepareEvaluationSets ---
set.seed(101)
n_seq <- 50
evaluationSummary <- list(
  seqIDs = paste0("Seq", 1:n_seq),
  tools = list(
    ToolX = sample(c(0, 1), n_seq, replace = TRUE),
    ToolY = sample(c(0, 1), n_seq, replace = TRUE),
    ToolZ = sample(c(0, 1), n_seq, replace = TRUE)
  ),
  type = sample(c("nc", "cds"), n_seq, replace = TRUE),
  isNC = sample(c(0, 1), n_seq, replace = TRUE)
)

# --- 2. Run the function ---
# Example 1: Analyze combinations of specific tools
results_comb <- evaluateToolCombinations(
  summaryList = evaluationSummary,
  tools = c("ToolX", "ToolY", "ToolZ")
)

if (!is.null(results_comb)) {
  print("Names of generated combinations:")
  print(names(results_comb$toolCombinations))

  print("Head of 'ToolX+ToolY' combination results:")
  print(head(results_comb$toolCombinations[["ToolX+ToolY"]]))
}

# Example 2: Non-interactively analyze all tools
results_all_comb <- evaluateToolCombinations(summaryList = evaluationSummary)

---

Evaluate Agreement Thresholds Among Coding Potential Tools

Description

Calculates agreement statistics for a selection of coding potential tools. For each sequence, it determines if it was predicted as non-coding by at least 'n' of the selected tools, for all possible values of 'n'.

Usage

evaluateToolsThresholds(summaryList, tools = NULL)

Arguments

summaryList	A list generated by summarizeTestSet(), containing ⁠$seqIDs⁠, ⁠$tools⁠, ⁠$isNC⁠, and ⁠$type⁠.
tools	An optional character vector specifying which tools to include in the agreement analysis. If NULL, the behavior depends on the session: interactive mode will prompt for selection, while non-interactive mode will use all available tools.

Value

A list containing:

seqIDs	Original sequence identifiers from the input.
isNC	Original true labels (1=nc, 0=cds) from the input.
type	Original type annotation ('nc', 'cds') from the input.
selectedToolsPredictions	A list containing only the prediction vectors for the tools that were selected for analysis.
sumsSelectedTools	An integer vector with the count of selected tools that predicted "non-coding" (1) for each sequence.
atLeastN	A list where each element ⁠atl{n}⁠ is a binary vector indicating if a sequence was predicted as non-coding by at least n selected tools. This list contains all threshold sets.

Examples

# --- 1. Create Example Data (mimicking summarizeTestSet output) ---
set.seed(456)
n_seq <- 50
exampleSummaryList <- list(
  seqIDs = paste0("ID", 1:n_seq),
  tools = list(
    ToolA = sample(c(0, 1), n_seq, replace = TRUE),
    ToolB = sample(c(0, 1), n_seq, replace = TRUE),
    ToolC = sample(c(0, 1), n_seq, replace = TRUE)
  ),
  type = sample(c("nc", "cds"), n_seq, replace = TRUE),
  isNC = sample(c(0, 1), n_seq, replace = TRUE)
)

# --- 2. Run the function ---
results <- evaluateToolsThresholds(
  summaryList = exampleSummaryList,
  tools = c("ToolA", "ToolB")
)

if (!is.null(results)) {
  # Accessing the nested list of thresholds
  print("Accessing the 'at-least-2' threshold vector:")
  print(head(results$atLeastN$atl2))
}

---

Find Potential cis-Regulatory Interactions from FEELnc Output

Description

This function reads and filters the output file from FEELnc (.classes file) to identify potential cis-interactions between lncRNAs and mRNAs based on user-specified criteria such as genomic distance and specific gene/transcript lists.

Usage

findCisInteractions(
  FEELncClassesFile,
  lncRnas = NULL,
  mRnas = NULL,
  filterIsBest = TRUE,
  lncRnaLevel = c("transcript", "gene"),
  mRnaLevel = c("gene", "transcript"),
  maxDist = 1e+05
)

Arguments

FEELncClassesFile	A character string specifying the path to the FEELnc .classes output file. This file is required.
lncRnas	An optional character vector of lncRNA IDs (gene or transcript) to filter the results. If NULL, no filtering by lncRNA ID is performed.
mRnas	An optional character vector of mRNA IDs (gene or transcript) to filter the results. If NULL, no filtering by mRNA ID is performed.
filterIsBest	Logical. If TRUE (default), only interactions marked as "best" (isBest == 1) in the FEELnc output are kept.
lncRnaLevel	A character string specifying the level for lncRNA filtering: "transcript" (default) or "gene".
mRnaLevel	A character string specifying the level for mRNA filtering: "gene" (default) or "transcript".
maxDist	A numeric value for the maximum distance (in base pairs) to consider for a cis-interaction (default: 100,000).

Value

A data.frame containing the filtered cis-interaction data with renamed columns (lncRNAId, targetRNAId).

Examples

# --- 1. Create a temporary FEELnc output file for a reproducible example ---
feelncFile <- tempfile()
mock_data <- data.frame(
  isBest = c(1, 1, 0, 1, 1),
  lncRNA_gene = c("LNC_G1", "LNC_G1", "LNC_G2", "LNC_G3", "LNC_G4"),
  lncRNA_transcript = c("LNC_T1", "LNC_T2", "LNC_T3", "LNC_T4", "LNC_T5"),
  partnerRNA_gene = c("TARGET_G1", "TARGET_G2", "TARGET_G1", "TARGET_G3", "TARGET_G4"),
  partnerRNA_transcript = c("T_T1", "T_T2", "T_T3", "T_T4", "T_T5"),
  distance = c(5000, 80000, 1000, 120000, 9000)
)
utils::write.table(mock_data, feelncFile, sep = "\t", row.names = FALSE, col.names = TRUE)

# --- 2. Define lncRNA and mRNA lists for filtering ---
lncRnaList <- c("LNC_T1", "LNC_T4") # Filter by transcript ID
mRnaList <- c("TARGET_G1")       # Filter by gene ID

# --- 3. Run the function ---
cis_interactions <- findCisInteractions(
  FEELncClassesFile = feelncFile,
  lncRnas = lncRnaList,
  mRnas = mRnaList,
  lncRnaLevel = "transcript",
  mRnaLevel = "gene",
  maxDist = 100000
)

print(cis_interactions)

# --- 4. Clean up the temporary file ---
unlink(feelncFile)

---

Estimates trans-interactions between lncRNAs and target RNAs based on expression correlation. This function requires the Hmisc and tidyr packages.

Description

Estimates trans-interactions between lncRNAs and target RNAs based on expression correlation. This function requires the Hmisc and tidyr packages.

Usage

findTransInteractions(
  exprMatrix,
  corMethod = "pearson",
  rval = 0.7,
  pval = 0.05,
  lncRnaList = NULL,
  tarRnaList = NULL,
  fullCorMatrixFile = NULL
)

Arguments

exprMatrix	A numeric matrix or data.frame of expression values. Rownames should contain gene/transcript IDs and columns should be samples.
corMethod	Correlation method: "pearson" (default) or "spearman".
rval	The cutoff for the correlation coefficient (default: 0.7).
pval	The cutoff for the p-value (default: 0.05).
lncRnaList	A list of lncRNA gene/transcript IDs. Must be present in rownames(exprMatrix). If NULL, all rownames are considered.
tarRnaList	A list of target gene/transcript IDs. Must be present in rownames(exprMatrix). If NULL, all rownames are considered.
fullCorMatrixFile	An optional file path to save the full correlation matrix.

Value

A data.frame of significant trans-interactions with columns for lncRNA ID, target RNA ID, r-value, and p-value.

Examples

# --- 1. Create a mock expression matrix ---
set.seed(123)
lnc_genes <- paste0("LNC", 1:5)
target_genes <- paste0("TARGET", 1:20)
all_genes <- c(lnc_genes, target_genes)
mockExprMatrix <- matrix(rnorm(25 * 10), nrow = 25, ncol = 10,
                         dimnames = list(all_genes, paste0("Sample", 1:10)))
mockExprMatrix["LNC1", ] <- mockExprMatrix["TARGET1", ] * 2 + rnorm(10, 0, 0.2)

# --- 2. Run the function ---
trans_interactions <- findTransInteractions(
  exprMatrix = mockExprMatrix,
  lncRnaList = lnc_genes,
  tarRnaList = target_genes,
  rval = 0.9,
  pval = 0.05
)
print(trans_interactions)

---

Extract Gene or Transcript Biotypes from a GRanges Object

Description

This function extracts unique biotypes for genes or transcripts from a GRanges object, typically imported from a reference GTF/GFF file.

Usage

getBiotypes(refGtf, level = "transcript")

Arguments

refGtf	A GRanges object imported from a reference annotation file. It must contain metadata columns for biotypes (e.g., gene_biotype, transcript_biotype) and corresponding IDs (gene_id, transcript_id).
level	A character string specifying whether to extract information for "gene" or "transcript" features (default: "transcript").

Value

A DataFrame (from the S4Vectors package) with two columns: the identifier (gene_id or transcript_id) and the corresponding biotype (gene_biotype or transcript_biotype).

Examples

# --- 1. Create a sample GRanges object mimicking a reference GTF ---
sampleRefGtf <- GenomicRanges::GRanges(
  seqnames = "chr1",
  ranges = IRanges::IRanges(start = 1:6, width = 100),
  gene_id = c("G1", "G1", "G2", "G2", "G3", "G3"),
  gene_biotype = c("protein_coding", "protein_coding", "lncRNA",
                   "lncRNA", "pseudogene", "pseudogene"),
  transcript_id = c("T1.1", "T1.2", "T2.1", "T2.1", "T3.1", "T3.1"),
  transcript_biotype = c("protein_coding", "protein_coding_variant",
                         "lncRNA", "lncRNA", "pseudogene", "pseudogene")
)

# --- 2. Extract biotypes at the transcript level ---
transcript_biotypes <- getBiotypes(refGtf = sampleRefGtf, level = "transcript")
print(transcript_biotypes)

# --- 3. Extract biotypes at the gene level ---
gene_biotypes <- getBiotypes(refGtf = sampleRefGtf, level = "gene")
print(gene_biotypes)

---

Extract Transcript Statistics from a GTF object

Description

This function takes a GRanges object (imported from a GTF file) and calculates the number of exons and total transcript length for each transcript.

Usage

getGtfStats(gtfObject)

Arguments

gtfObject

A GRanges object, typically imported from a GTF file using rtracklayer::import().

Value

A data.frame with columns: "transcript_id", "exons", and "trans_length".

Examples

# Create a sample GRanges object to mimic a GTF import
sample_gtf <- GenomicRanges::GRanges(
  seqnames = "chr1",
  ranges = IRanges::IRanges(
    start = c(100, 300, 800, 950),
    end = c(200, 400, 900, 1050)
  ),
  strand = "+",
  type = "exon",
  transcript_id = c("T1", "T1", "T2", "T2"),
  exon_number = c("1", "2", "1", "20") # Example with exon_number > 9
)

# Calculate statistics
transcript_stats <- getGtfStats(sample_gtf)
print(transcript_stats)

---

Create Clock Plots for Confusion Matrix Metrics

Description

Generates circular bar plots (clock plots) to visualize performance metrics. Supports both single and multiple plot layouts for comparing methods.

Usage

plotClockMetrics(
  cmList,
  methods = NULL,
  metrics = c("Accuracy", "Sensitivity", "Specificity", "Precision", "Recall"),
  plotTitle = "Clock Plot of Metrics",
  colors = NULL,
  layout = c("single", "multiple"),
  ...
)

Arguments

cmList	A named list where each element represents a confusion matrix, typically the output from calculateCM.
methods	An optional character vector of method names to include. If NULL, interactive selection is triggered in an interactive session; otherwise, all methods are used.
metrics	A character vector of metrics to display. Defaults to c("Accuracy", "Sensitivity", "Specificity", "Precision", "Recall").
plotTitle	The title for the plot(s).
colors	An optional vector of colors for each method.
layout	The layout of the plots: "single" (default) for one plot, or "multiple" for a grid of plots.
...	Additional arguments (not currently used).

Value

A ggplot object (for layout = "single") or a patchwork object (for layout = "multiple"), which can be printed to display.

Examples

# --- 1. Create a mock cmList object (as from calculateCM) ---
set.seed(123)
mockCmList <- list(
  `MethodA` = list(metrics = c(Accuracy=0.9, Sensitivity=0.8, Specificity=0.95,
                               Precision=0.85, Recall=0.8)),
  `MethodB` = list(metrics = c(Accuracy=0.8, Sensitivity=0.9, Specificity=0.7,
                               Precision=0.75, Recall=0.9))
)

# --- 2. Run the plot function ---
# Example 1: Single plot comparing selected methods
plotClockMetrics(
  cmList = mockCmList,
  methods = c("MethodA", "MethodB"),
  plotTitle = "Comparison Clock Plot"
)

# Example 2: Multiple plots, one for each method
plotClockMetrics(cmList = mockCmList, layout = "multiple")

---

Create Radar Plots for Confusion Matrix Metrics

Description

Generates radar plots to visualize and compare performance metrics for different methods or tool combinations. Supports a single plot for comparing multiple methods or a grid of plots for individual evaluation.

Usage

plotRadarMetrics(
  cmList,
  methods = NULL,
  metrics = c("Accuracy", "Sensitivity", "Specificity", "Precision", "Recall"),
  plotTitle = "Radar Plot of Metrics",
  colors = NULL,
  layout = c("single", "multiple"),
  displayArea = FALSE,
  displayFill = TRUE,
  saveData = FALSE,
  fileName = NULL,
  ...
)

Arguments

cmList	A named list where each element represents a confusion matrix, typically the output from calculateCM. Each element must be a list containing at least a named numeric vector called ⁠$metrics⁠.
methods	An optional character vector of method names (from names(cmList)) to include in the plot. If NULL, behavior depends on the session: interactive mode prompts for selection, while non-interactive mode uses all available methods.
metrics	A character vector of metrics to visualize on the radar plot axes. Defaults to c("Accuracy", "Sensitivity", "Specificity", "Precision", "Recall").
plotTitle	A character string for the plot title.
colors	An optional character vector of colors for each method. If NULL, default colors are generated.
layout	Specifies the plot layout: "single" (default) for one plot comparing all methods, or "multiple" for a grid of individual plots.
displayArea	Logical. If TRUE, displays the calculated polygon area.
displayFill	Logical. If TRUE (default), fills radar chart polygons with a semi-transparent color.
saveData	Logical. If TRUE, saves the underlying data frame to a CSV file.
fileName	The name of the file for saving data (required if saveData is TRUE).
...	Additional arguments passed to fmsb::radarchart.

Value

This function is called for its side effect of generating plots and does not return a value (invisible(NULL)).

Examples

# --- 1. Create a mock cmList object (as from calculateCM) ---
set.seed(123)
mockCmList <- list(
  `MethodA` = list(metrics = c(Accuracy=0.9, Sensitivity=0.8, Specificity=0.95,
                               Precision=0.85, Recall=0.8)),
  `MethodB` = list(metrics = c(Accuracy=0.8, Sensitivity=0.9, Specificity=0.7,
                               Precision=0.75, Recall=0.9)),
  `MethodC` = list(metrics = c(Accuracy=0.85, Sensitivity=0.85, Specificity=0.85,
                               Precision=0.85, Recall=0.85))
)

# --- 2. Run the plot function ---
# To prevent plots from showing up during automated checks, we wrap in a device
temp_png <- tempfile(fileext = ".png")
png(temp_png)

# Example 1: Single plot comparing selected methods
plotRadarMetrics(
  cmList = mockCmList,
  methods = c("MethodA", "MethodC"),
  plotTitle = "Comparison Plot"
)

# Example 2: Multiple plots, one for each method
plotRadarMetrics(cmList = mockCmList, layout = "multiple")

dev.off()
unlink(temp_png)

---

Plot Functional Interactions as a Sankey Diagram

Description

Visualizes genomic interaction data (lncRNAId -> Target -> Functional Term) as an interactive Sankey diagram using Plotly. Replaces plot_by_terms, plot_by_lnc, etc., providing a unified interface for filtering and plotting.

Usage

plotSankeyInteractions(
  interactionData,
  groupBy = NULL,
  selectIds = NULL,
  showLabels = FALSE,
  highlightSelected = FALSE,
  color = NULL,
  title = NULL
)

Arguments

interactionData	A data.frame containing interaction data, typically created by annotateInteractions(). Must contain columns: "lncRNAId", "intersection", "term_name", and "type".
groupBy	A character string specifying the filtering criteria. Valid options are: "lncRNAId", "target", "term", "type". If NULL, interactive selection is triggered.
selectIds	Optional character vector of IDs/terms to filter or highlight. If NULL and interactive, prompts for selection.
showLabels	Logical. If TRUE, displays labels on nodes. Defaults to FALSE.
highlightSelected	Logical. If TRUE, plots all data but colors only the nodes/links associated with selectIds. If FALSE (default), plots only the data matching selectIds.
color	Optional character string. A single color for all nodes/links. If NULL, a random palette (Polychrome) is generated.
title	Optional character string for the plot title.

Value

A plotly object containing the Sankey diagram.

Examples

# --- 1. Create mock interaction data ---
mockData <- data.frame(
  lncRNAId = c(rep("LNC1", 3), rep("LNC2", 2)),
  intersection = c("T1", "T2", "T1", "T3", "T2"),
  term_name = c("Stress", "Growth", "Stress", "Immunity", "Growth"),
  type = c("cis", "cis", "trans", "trans", "cis"),
  stringsAsFactors = FALSE
)

# --- 2. Run in non-interactive mode ---

# Example 1: Filter by specific Term
fig1 <- plotSankeyInteractions(
  interactionData = mockData,
  groupBy = "term",
  selectIds = "Stress",
  title = "Stress Interactions"
)
# fig1

# Example 2: Highlight specific lncRNAId
fig2 <- plotSankeyInteractions(
  interactionData = mockData,
  groupBy = "lncRNAId",
  selectIds = "LNC1",
  highlightSelected = TRUE,
  title = "Highlighting LNC1"
)
# fig2

---

Create Venn Diagram from Coding Potential Results

Description

Generates a Venn diagram of noncoding predictions from multiple tools, based on the output of aggregateCodPot. This function requires the 'venn' package, which should be declared in the Suggests field of the DESCRIPTION file.

Usage

plotVennCodPot(
  codPot,
  selection = NULL,
  vennColors = grDevices::palette.colors(n = 9, palette = "Okabe-Ito")
)

Arguments

codPot	A list object generated by aggregateCodPot().
selection	An optional logical vector indicating which tools to include (TRUE for include, FALSE for exclude). If NULL, all available tools are used.
vennColors	A vector of colors for the Venn diagram segments.

Value

Invisibly returns the result of venn::venn. The primary effect is plotting the diagram to the active graphics device.

Examples

# Example with all tools

  mockCodPot <- list(
  seqIDs = c("tx1", "tx2", "tx3", "tx4"),
  tools = list(
    CPC2 = c(0, 1, 1, 0),
    PLEK = c(0, 1, 0, 1)
  )
)
if (requireNamespace("venn", quietly = TRUE)) {
  plotVennCodPot(codPot = mockCodPot)
}

---

Prepare Data Sets for Performance Evaluation

Description

Filters coding potential results to include only sequences present in provided test sets. It annotates each sequence as non-coding (nc) or protein-coding (cds), creating a summary object ready for various evaluation functions (e.g., bestTool, bestToolAtleast).

Usage

prepareEvaluationSets(codPotList, ncTest, cdsTest)

Arguments

codPotList	A list object, typically from aggregateCodPot(). Must contain ⁠$seqIDs⁠ (character vector) and ⁠$tools⁠ (a list of numeric vectors).
ncTest	A character vector of sequence IDs known to be non-coding.
cdsTest	A character vector of sequence IDs known to be protein-coding.

Value

A list containing elements filtered to include only sequences from ncTest or cdsTest:

seqIDs	Filtered character vector of sequence IDs.
tools	Filtered list of tool prediction vectors.
type	Character vector with annotation ("nc" or "cds").
isNC	Integer vector: 1 if type is "nc", 0 if "cds".
sums	Integer vector with the sum of predictions across all tools.

Examples

# --- 1. Create mock data mimicking package outputs ---
# a) Output from aggregateCodPot()
mockCodPotList <- list(
  seqIDs = c("nc_seq1", "cds_seq1", "nc_seq2", "other_seq"),
  tools = list(
    CPC2 = c(1, 0, 1, 1),
    PLEK = c(1, 1, 0, 0)
  )
)

# b) Outputs from createTrainTestSets()
mockNcSets <- list(
  nc.train = c("nc_seq3", "nc_seq4"),
  nc.test = c("nc_seq1", "nc_seq2")
)
mockCdsSets <- list(
  cds.train = c("cds_seq2"),
  cds.test = c("cds_seq1")
)

# --- 2. Run the function to prepare the evaluation set ---
evaluationSummary <- prepareEvaluationSets(
  codPotList = mockCodPotList,
  ncTest = mockNcSets$nc.test,
  cdsTest = mockCdsSets$cds.test
)

# --- 3. Inspect the prepared data ---
# Note: "other_seq" was filtered out as it was not in the test sets.
print(evaluationSummary)

---