Package 'LBDiscover'

Description

This function implements the ABC model for literature-based discovery with enhanced term filtering and validation.

Usage

abc_model(
  co_matrix,
  a_term,
  c_term = NULL,
  min_score = 0.1,
  n_results = 100,
  scoring_method = c("multiplication", "average", "combined", "jaccard"),
  b_term_types = NULL,
  c_term_types = NULL,
  exclude_general_terms = TRUE,
  filter_similar_terms = TRUE,
  similarity_threshold = 0.8,
  enforce_strict_typing = TRUE,
  validation_method = "pattern"
)

Arguments

Value

A data frame with ranked discovery results.

Examples

# Create a simple example co-occurrence matrix
set.seed(123)
terms <- c("migraine", "headache", "pain", "serotonin", "sumatriptan", "CGRP")
n_terms <- length(terms)
co_matrix <- matrix(runif(n_terms^2, 0, 1), nrow = n_terms, ncol = n_terms)
rownames(co_matrix) <- colnames(co_matrix) <- terms
diag(co_matrix) <- 1  # Perfect self-similarity

# Add entity types
entity_types <- c("disease", "symptom", "symptom", "chemical", "drug", "protein")
names(entity_types) <- terms
attr(co_matrix, "entity_types") <- entity_types

# Apply ABC model
results <- abc_model(co_matrix, a_term = "migraine", min_score = 0.1, n_results = 5)
print(results)

Description

This function implements an optimized version of the ABC model calculation that's more efficient for large co-occurrence matrices.

Usage

abc_model_opt(
  co_matrix,
  a_term,
  c_term = NULL,
  min_score = 0.1,
  n_results = 100,
  chunk_size = 500
)

Arguments

Value

A data frame with ranked discovery results.

Examples

## Not run: 
abc_results <- abc_model_opt(co_matrix, a_term = "migraine")

## End(Not run)

Description

This function extends the ABC model with statistical significance testing to evaluate the strength of discovered connections.

Usage

abc_model_sig(
  co_matrix,
  a_term,
  c_term = NULL,
  a_type = NULL,
  c_type = NULL,
  min_score = 0.1,
  n_results = 100,
  n_permutations = 1000,
  scoring_method = c("multiplication", "average", "combined", "jaccard")
)

Arguments

Value

A data frame with ranked discovery results and p-values.

Examples

## Not run: 
abc_results <- abc_model_sig(co_matrix, a_term = "migraine",
                                                scoring_method = "combined")

## End(Not run)

Description

This function implements a time-sliced ABC model for validation. It uses historical data to predict connections that will appear in the future.

Usage

abc_timeslice(
  entity_data,
  time_column = "publication_year",
  split_time,
  a_term,
  a_type = NULL,
  c_type = NULL,
  min_score = 0.1,
  n_results = 100
)

Arguments

Value

A list with prediction results and validation metrics.

Examples

## Not run: 
validation <- abc_timeslice(entity_data,
                            time_column = "publication_year",
                            split_time = 2010,
                            a_term = "migraine")

## End(Not run)

Description

This function implements an improved ANC model that ensures only biomedical terms are used as intermediaries.

Usage

anc_model(
  co_matrix,
  a_term,
  n_b_terms = 3,
  c_type = NULL,
  min_score = 0.1,
  n_results = 100,
  enforce_biomedical_terms = TRUE,
  b_term_types = c("protein", "gene", "chemical", "pathway", "drug", "disease",
    "biological_process"),
  validation_function = is_valid_biomedical_entity
)

Arguments

Value

A data frame with ranked discovery results.

Description

This function implements a BITOLA-style discovery model based on MeSH term co-occurrence and semantic type filtering.

Usage

bitola_model(
  co_matrix,
  a_term,
  a_semantic_type = NULL,
  c_semantic_type = NULL,
  min_score = 0.1,
  n_results = 100
)

Arguments

Value

A data frame with ranked discovery results.

Examples

## Not run: 
bitola_results <- bitola_model(co_matrix, a_term = "migraine",
                                   a_semantic_type = "Disease",
                                   c_semantic_type = "Gene")

## End(Not run)

Description

This function calculates basic bibliometric statistics from article data.

Usage

calc_bibliometrics(article_data, by_year = TRUE)

Arguments

Value

A list containing bibliometric statistics.

Examples

## Not run: 
stats <- calc_bibliometrics(article_data)

## End(Not run)

Description

This function calculates the similarity between documents using TF-IDF weighting and cosine similarity.

Usage

calc_doc_sim(
  text_data,
  text_column = "abstract",
  min_term_freq = 2,
  max_doc_freq = 0.9
)

Arguments

Value

A similarity matrix for the documents.

Examples

## Not run: 
sim_matrix <- calc_doc_sim(article_data, text_column = "abstract")

## End(Not run)

Description

Removes all cached PubMed search results

Usage

clear_pubmed_cache()

Value

NULL invisibly

Examples

# Clear the PubMed cache (always safe to run)
clear_pubmed_cache()

Description

This function clusters documents using K-means based on their TF-IDF vectors.

Usage

cluster_docs(
  text_data,
  text_column = "abstract",
  n_clusters = 5,
  min_term_freq = 2,
  max_doc_freq = 0.9,
  random_seed = 42
)

Arguments

Value

A data frame with the original data and cluster assignments.

Examples

## Not run: 
clustered_data <- cluster_docs(article_data, text_column = "abstract", n_clusters = 5)

## End(Not run)

Description

This function compares term frequencies between two sets of articles.

Usage

compare_terms(
  corpus1,
  corpus2,
  text_column = "abstract",
  corpus1_name = "Corpus1",
  corpus2_name = "Corpus2",
  n = 100,
  remove_stopwords = TRUE
)

Arguments

Value

A data frame containing term frequency comparisons.

Examples

## Not run: 
comparison <- compare_terms(corpus1, corpus2,
                                      corpus1_name = "Migraine",
                                      corpus2_name = "Magnesium")

## End(Not run)

Description

This function creates a citation network from article data. Note: Currently a placeholder as it requires citation data not available through basic PubMed queries.

Usage

create_citation_net(article_data, citation_data = NULL)

Arguments

Value

An igraph object representing the citation network.

Examples

## Not run: 
network <- create_citation_net(article_data)

## End(Not run)

Description

This function creates a co-occurrence matrix from entity data while preserving entity type information as an attribute without enforcing type constraints.

Usage

create_comat(
  entity_data,
  doc_id_col = "doc_id",
  entity_col = "entity",
  count_col = NULL,
  type_col = "entity_type",
  normalize = TRUE,
  normalization_method = c("cosine", "jaccard", "dice")
)

Arguments

Value

A co-occurrence matrix with entity types stored as an attribute.

Examples

## Not run: 
co_matrix <- create_comat(entities,
                          doc_id_col = "doc_id",
                          entity_col = "entity",
                          type_col = "entity_type")

## End(Not run)

Description

This function generates an HTML report summarizing discovery results without enforcing entity type constraints. It includes data validation to avoid errors with publication years and other data issues.

Usage

create_report(
  results,
  visualizations = NULL,
  articles = NULL,
  output_file = "discovery_report.html"
)

Arguments

Value

The file path of the created HTML report (invisibly).

Description

This function creates a sparse co-occurrence matrix from entity data, which is more memory-efficient for large datasets.

Usage

create_sparse_comat(
  entity_data,
  doc_id_col = "doc_id",
  entity_col = "entity",
  count_col = NULL,
  type_col = NULL,
  normalize = TRUE
)

Arguments

Value

A sparse matrix of entity co-occurrences.

Examples

## Not run: 
co_matrix <- create_sparse_comat(entities,
                                              doc_id_col = "doc_id",
                                              entity_col = "entity")

## End(Not run)

Description

This function creates a term-document matrix from preprocessed text data.

Usage

create_tdm(preprocessed_data, min_df = 2, max_df = 0.9)

Arguments

Value

A term-document matrix.

Examples

## Not run: 
preprocessed <- preprocess_text(article_data, text_column = "abstract")
tdm <- create_tdm(preprocessed)

## End(Not run)

Description

This function creates a term-document matrix from preprocessed text data. It's a simplified version of create_tdm() for direct use in models.

Usage

create_term_document_matrix(preprocessed_data, min_df = 2, max_df = 0.9)

Arguments

Value

A term-document matrix.

Examples

## Not run: 
preprocessed <- preprocess_text(article_data, text_column = "abstract")
tdm <- create_term_document_matrix(preprocessed)

## End(Not run)

Description

This function attempts to detect the language of a text string. It implements a simple n-gram based approach that doesn't require additional packages.

Usage

detect_lang(text, sample_size = 1000)

Arguments

Value

Character string containing the ISO 639-1 language code

Examples

## Not run: 
lang <- detect_lang("This is English text")
# Returns "en"

## End(Not run)

Description

This function applies diversity enforcement to ABC model results by:

1. Removing duplicate paths to the same C term

2. Ensuring B term diversity by selecting top results from each B term group

3. Preventing A and C terms from appearing as B terms

Usage

diversify_abc(
  abc_results,
  diversity_method = c("both", "b_term_groups", "unique_c_paths"),
  max_per_group = 3,
  min_score = 0.1
)

Arguments

Value

A data frame with diverse ABC results.

Examples

## Not run: 
diverse_results <- diversify_abc(abc_results)

## End(Not run)

Description

This function enhances ABC results with information from external knowledge bases.

Usage

enhance_abc_kb(abc_results, knowledge_base = c("umls", "mesh"), api_key = NULL)

Arguments

Value

A data frame with enhanced ABC results.

Examples

## Not run: 
enhanced_results <- enhance_abc_kb(abc_results, knowledge_base = "mesh")

## End(Not run)

Description

This function evaluates the top results by searching for supporting evidence in the literature for the connections.

Usage

eval_evidence(
  results,
  max_results = 5,
  base_term = NULL,
  max_articles = 5,
  verbose = TRUE
)

Arguments

Value

A list containing evaluation results

Examples

# Create example results for evaluation
results <- data.frame(
  a_term = rep("migraine", 2),
  b_term = c("serotonin", "CGRP"),
  c_term = c("sumatriptan", "fremanezumab"),
  abc_score = c(0.8, 0.7),
  c_type = rep("drug", 2)
)

## Not run: 
# This requires internet connection for PubMed search
evaluation <- eval_evidence(results, max_results = 2, base_term = "migraine")
print(names(evaluation))

## End(Not run)

Description

This function creates an HTML chord diagram visualization for ABC connections.

Usage

export_chord(
  abc_results,
  output_file = "abc_chord.html",
  top_n = 50,
  min_score = 0.1,
  open = TRUE
)

Arguments

Value

The file path of the created HTML file (invisibly).

Examples

## Not run: 
export_chord(abc_results, output_file = "abc_chord.html")

## End(Not run)

Description

This function creates an HTML chord diagram visualization for ABC connections, properly coloring the arcs based on whether each term is an A, B, or C term.

Usage

export_chord_diagram(
  abc_results,
  output_file = "abc_chord.html",
  top_n = 50,
  min_score = 0.1,
  open = TRUE
)

Arguments

Value

The file path of the created HTML file (invisibly).

Examples

## Not run: 
export_chord_diagram(abc_results, output_file = "abc_chord.html")

## End(Not run)

Description

This function exports ABC results to a simple HTML file with a visualization. If the visNetwork package is available, it will use it for a more interactive visualization.

Usage

export_network(
  abc_results,
  output_file = "abc_network.html",
  top_n = 50,
  min_score = 0.1,
  open = TRUE
)

Arguments

Value

The file path of the created HTML file (invisibly).

Examples

## Not run: 
export_network(abc_results, output_file = "abc_network.html")

## End(Not run)

Description

This function extracts entities from text and optionally assigns them to specific semantic categories based on dictionaries.

Usage

extract_entities(
  text_data,
  text_column = "abstract",
  dictionary = NULL,
  case_sensitive = FALSE,
  overlap_strategy = c("priority", "all", "longest"),
  sanitize_dict = TRUE
)

Arguments

Value

A data frame with extracted entities, their types, and positions.

Examples

# Create example text data
text_data <- data.frame(
  doc_id = 1:3,
  abstract = c(
    "Migraine is a neurological disorder causing severe headache and photophobia.",
    "Serotonin receptors play a role in migraine pathophysiology.",
    "Sumatriptan is an effective treatment for migraine attacks."
  )
)

# Create example dictionary
dictionary <- data.frame(
  term = c("migraine", "headache", "photophobia", "serotonin", "sumatriptan"),
  type = c("disease", "symptom", "symptom", "chemical", "drug")
)

# Extract entities
entities <- extract_entities(text_data, dictionary = dictionary)
print(entities)

Description

This function provides a complete workflow for extracting entities from text using dictionaries from multiple sources, with improved performance and robust error handling.

Usage

extract_entities_workflow(
  text_data,
  text_column = "abstract",
  entity_types = c("disease", "drug", "gene"),
  dictionary_sources = c("local", "mesh", "umls"),
  additional_mesh_queries = NULL,
  sanitize = TRUE,
  api_key = NULL,
  custom_dictionary = NULL,
  max_terms_per_type = 200,
  verbose = TRUE,
  batch_size = 500,
  parallel = FALSE,
  num_cores = 2,
  cache_dictionaries = TRUE
)

Arguments

Value

A data frame with extracted entities, their types, and positions.

Examples

# Create example text data
text_data <- data.frame(
  doc_id = 1:2,
  abstract = c(
    "Migraine is a neurological disorder.",
    "Serotonin plays a role in headache."
  )
)

# Extract entities using workflow
entities <- extract_entities_workflow(
  text_data,
  entity_types = c("disease", "chemical"),
  dictionary_sources = "local",
  max_terms_per_type = 10
)
print(head(entities))

Description

This function performs a simple dictionary-based named entity recognition. For more advanced NER, consider using external tools via reticulate.

Usage

extract_ner(
  text,
  entity_types = c("disease", "drug", "gene"),
  custom_dictionaries = NULL
)

Arguments

Value

A data frame containing found entities, their types, and positions

Examples

## Not run: 
entities <- extract_ner(abstracts,
                                  entity_types = c("disease", "drug", "gene"))

## End(Not run)

Description

This function extracts n-grams (sequences of n words) from text.

Usage

extract_ngrams(text, n = 1, min_freq = 2)

Arguments

Value

A data frame containing n-grams and their frequencies

Examples

## Not run: 
bigrams <- extract_ngrams(abstracts, n = 2)

## End(Not run)

Description

This function extracts and counts the most common terms in a corpus.

Usage

extract_terms(
  article_data,
  text_column = "abstract",
  n = 100,
  remove_stopwords = TRUE,
  min_word_length = 3
)

Arguments

Value

A data frame containing term counts.

Examples

## Not run: 
common_terms <- extract_terms(article_data, text_column = "abstract")

## End(Not run)

Description

This function implements a simple non-negative matrix factorization (NMF) approach to topic modeling, without requiring additional packages.

Usage

extract_topics(
  text_data,
  text_column = "abstract",
  n_topics = 5,
  max_terms = 10,
  n_iterations = 50
)

Arguments

Value

A list containing topic-term and document-topic matrices

Examples

## Not run: 
topics <- extract_topics(article_data, text_column = "abstract", n_topics = 5)

## End(Not run)

Description

Filter a co-occurrence matrix by entity type

Usage

filter_by_type(co_matrix, types)

Arguments

Value

A filtered co-occurrence matrix.

Examples

## Not run: 
# Keep only disease and drug entities
filtered_matrix <- filter_by_type(co_matrix, types = c("disease", "drug"))

## End(Not run)

Description

This function finds all potential ABC connections in a co-occurrence matrix.

Usage

find_abc_all(
  co_matrix,
  a_type = NULL,
  c_type = NULL,
  min_score = 0.1,
  n_results = 1000
)

Arguments

Value

A data frame with ranked discovery results.

Examples

## Not run: 
all_abc <- find_abc_all(co_matrix, a_type = "source", c_type = "target")

## End(Not run)

Description

This function finds documents similar to a given document based on TF-IDF and cosine similarity.

Usage

find_similar_docs(text_data, doc_id, text_column = "abstract", n_similar = 5)

Arguments

Value

A data frame with similar documents and their similarity scores.

Examples

## Not run: 
similar_docs <- find_similar_docs(article_data, doc_id = 1,
                                     text_column = "abstract", n_similar = 5)

## End(Not run)

Description

This function verifies that the primary term exists in the co-occurrence matrix, and if not, attempts to find a suitable variation.

Usage

find_term(co_matrix, primary_term, verbose = TRUE)

Arguments

Value

The found term (either exact match or variation)

Examples

# Create example co-occurrence matrix
terms <- c("migraine", "headache", "pain", "serotonin")
co_matrix <- matrix(runif(16, 0, 1), nrow = 4, ncol = 4)
rownames(co_matrix) <- colnames(co_matrix) <- terms

# Find term in matrix
found_term <- find_term(co_matrix, "migraine")
print(found_term)

Description

This function creates a comprehensive HTML report from discovery results and visualizations.

Usage

gen_report(
  results_list,
  visualizations = NULL,
  articles = NULL,
  output_file = "discoveries.html",
  verbose = TRUE
)

Arguments

Value

Invisible output_file path

Examples

# Create example data for report generation
results_list <- list(
  abc_results = data.frame(
    a_term = "migraine",
    c_term = "sumatriptan",
    abc_score = 0.8
  )
)


# Generate report to temporary file
temp_file <- tempfile(fileext = ".html")
gen_report(results_list, output_file = temp_file)

Description

Get dictionary cache environment

Usage

get_dict_cache()

Value

The environment containing cached dictionary data

Examples

# Get the dictionary cache environment
cache_env <- get_dict_cache()
print(ls(cache_env))

Description

This function retrieves full text articles from PubMed Central.

Usage

get_pmc_fulltext(pmids, api_key = NULL)

Arguments

Value

A data frame containing article metadata and full text.

Examples

## Not run: 
full_texts <- get_pmc_fulltext(c("12345678", "23456789"))

## End(Not run)

Description

This function identifies variations of a primary term within a corpus of articles.

Usage

get_term_vars(articles, primary_term, text_col = "abstract")

Arguments

Value

A character vector of unique term variations, sorted by length

Examples

# Create example articles
articles <- data.frame(
  abstract = c(
    "Migraine headaches are debilitating",
    "Migraines affect quality of life",
    "Migraine disorders require treatment"
  )
)

# Get term variations
variations <- get_term_vars(articles, "migrain")
print(variations)

Description

Get entity type distribution from co-occurrence matrix

Usage

get_type_dist(co_matrix)

Arguments

Value

A data frame with entity type counts and percentages.

Examples

## Not run: 
type_dist <- get_type_dist(co_matrix)

## End(Not run)

Description

Determine if a term is likely a specific biomedical entity with improved accuracy

Usage

is_valid_biomedical_entity(term, claimed_type = NULL)

Arguments

Value

Logical, TRUE if the term is likely a valid biomedical entity, FALSE otherwise

Examples

# Test biomedical entity validation
is_valid_biomedical_entity("migraine", "disease")  # Should return TRUE
is_valid_biomedical_entity("receptor", "protein")  # Should return TRUE
is_valid_biomedical_entity("optimization", "chemical")  # Should return FALSE
is_valid_biomedical_entity("europe", "disease")  # Should return FALSE

Description

This function loads pre-defined biomedical dictionaries or fetches terms from MeSH/UMLS.

Usage

load_dictionary(
  dictionary_type = NULL,
  custom_path = NULL,
  source = c("local", "mesh", "umls"),
  api_key = NULL,
  n_terms = 200,
  mesh_query = NULL,
  semantic_type_filter = NULL,
  sanitize = TRUE,
  extended_mesh = FALSE,
  mesh_queries = NULL
)

Arguments

Value

A data frame containing the dictionary.

Examples

# Load a disease dictionary from local source
disease_dict <- load_dictionary("disease", source = "local")
head(disease_dict)

# Load with custom terms
custom_dict <- data.frame(
  term = c("migraine", "headache", "photophobia"),
  type = c("disease", "symptom", "symptom"),
  id = c("D001", "D002", "D003"),
  source = rep("custom", 3)
)
print(custom_dict)

Description

This function loads previously saved results from a file.

Usage

load_results(file_path)

Arguments

Value

A data frame containing the loaded results.

Examples

## Not run: 
results <- load_results("search_results.csv")

## End(Not run)

Description

This function implements an improved LSI model that more rigorously filters out non-biomedical terms from the results to ensure clinical relevance.

Usage

lsi_model(
  term_doc_matrix,
  a_term,
  n_factors = 100,
  n_results = 100,
  enforce_biomedical_terms = TRUE,
  c_term_types = NULL,
  entity_types = NULL,
  validation_function = is_valid_biomedical_entity,
  min_word_length = 3,
  use_nlp = TRUE,
  nlp_threshold = 0.7
)

Arguments

Value

A data frame with ranked discovery results.

Description

This function maps terms to standard biomedical ontologies like MeSH or UMLS.

Usage

map_ontology(
  terms,
  ontology = c("mesh", "umls"),
  api_key = NULL,
  fuzzy_match = FALSE,
  similarity_threshold = 0.8,
  mesh_query = NULL,
  semantic_types = NULL,
  dictionary_type = "disease"
)

Arguments

Value

A data frame with mapped terms and ontology identifiers

Examples

## Not run: 
# Map terms to MeSH
mesh_mappings <- map_ontology(c("headache", "migraine"), ontology = "mesh")

# Map terms to UMLS with API key
umls_mappings <- map_ontology(c("headache", "migraine"), ontology = "umls",
                              api_key = "your_api_key")

## End(Not run)

Description

This function combines custom and standard entity datasets, handling the case where one or both might be empty, and removes duplicates.

Usage

merge_entities(
  custom_entities,
  standard_entities,
  primary_term,
  primary_type = "disease",
  verbose = TRUE
)

Arguments

Value

A data frame of combined entities

Examples

# Create example entity datasets
custom_entities <- data.frame(
  doc_id = c(1, 1, 2),
  entity = c("migraine", "headache", "pain"),
  entity_type = c("disease", "symptom", "symptom"),
  start_pos = c(1, 10, 5),
  end_pos = c(8, 18, 9),
  sentence = c("sent1", "sent1", "sent2"),
  frequency = c(2, 1, 1)
)

standard_entities <- data.frame(
  doc_id = c(1, 2, 2),
  entity = c("serotonin", "migraine", "therapy"),
  entity_type = c("chemical", "disease", "treatment"),
  start_pos = c(20, 1, 15),
  end_pos = c(29, 8, 22),
  sentence = c("sent1", "sent2", "sent2"),
  frequency = c(1, 1, 1)
)

# Merge entities
merged <- merge_entities(custom_entities, standard_entities, "migraine")
print(merged)

Description

This function merges multiple search results into a single data frame.

Usage

merge_results(..., remove_duplicates = TRUE)

Arguments

Value

A merged data frame.

Examples

## Not run: 
merged_results <- merge_results(results1, results2, results3)

## End(Not run)

Description

This function ensures there are sufficient results for visualization, creating placeholder data if necessary.

Usage

min_results(
  diverse_results,
  top_results,
  a_term,
  min_results = 3,
  fallback_count = 15,
  verbose = TRUE
)

Arguments

Value

A data frame with sufficient results for visualization

Examples

# Create example diverse results (empty case)
diverse_results <- data.frame()
top_results <- data.frame(
  a_term = rep("migraine", 3),
  b_term = c("serotonin", "CGRP", "cortisol"),
  c_term = c("sumatriptan", "fremanezumab", "propranolol"),
  abc_score = c(0.8, 0.7, 0.6)
)

# Ensure minimum results
final_results <- min_results(diverse_results, top_results, "migraine")
print(final_results)

Description

This function searches various NCBI databases using the E-utilities API via the rentrez package.

Usage

ncbi_search(
  query,
  database = "pubmed",
  max_results = 1000,
  use_mesh = FALSE,
  date_range = NULL,
  api_key = NULL,
  retry_count = 3,
  retry_delay = 2
)

Arguments

Value

A data frame containing the search results with IDs, titles, and other metadata.

Examples

## Not run: 
results <- ncbi_search("migraine headache", database = "pubmed", max_results = 100)
gene_results <- ncbi_search("BRCA1", database = "gene", max_results = 10)

## End(Not run)

Description

This function uses parallel processing to analyze documents faster.

Usage

parallel_analysis(
  text_data,
  analysis_function,
  text_column = "abstract",
  ...,
  n_cores = NULL
)

Arguments

Value

A data frame with analysis results.

Examples

## Not run: 
# Define a simple analysis function
count_words <- function(text) {
  words <- unlist(strsplit(tolower(text), "\\s+"))
  return(length(words))
}

# Apply parallel processing
results <- parallel_analysis(article_data, count_words, text_column = "abstract")

## End(Not run)

Description

This function assesses the significance of ABC model results through randomization. It generates a null distribution by permuting the co-occurrence matrix.

Usage

perm_test_abc(abc_results, co_matrix, n_permutations = 1000, alpha = 0.05)

Arguments

Value

A data frame with ABC results and permutation-based significance measures.

Examples

## Not run: 
randomized_results <- perm_test_abc(abc_results, co_matrix, n_permutations = 1000)

## End(Not run)

Description

This function creates a heatmap visualization from ABC results.

Usage

plot_heatmap(
  results,
  output_file = "heatmap.png",
  width = 1200,
  height = 900,
  resolution = 120,
  top_n = 15,
  min_score = 1e-04,
  color_palette = "blues",
  show_entity_types = TRUE,
  verbose = TRUE
)

Arguments

Value

Invisible NULL (creates a file as a side effect)

Examples

# Create example results for visualization
results <- data.frame(
  a_term = rep("migraine", 4),
  b_term = c("serotonin", "CGRP", "cortisol", "dopamine"),
  c_term = c("sumatriptan", "fremanezumab", "propranolol", "amitriptyline"),
  abc_score = c(0.8, 0.7, 0.6, 0.5),
  b_type = c("chemical", "protein", "hormone", "chemical"),
  c_type = rep("drug", 4)
)


# These require graphics capabilities
plot_heatmap(results, output_file = tempfile(fileext = ".png"))
plot_network(results, output_file = tempfile(fileext = ".png"))

Description

This function creates a network visualization from ABC results.

Usage

plot_network(
  results,
  output_file = "network.png",
  width = 1200,
  height = 900,
  resolution = 120,
  top_n = 15,
  min_score = 1e-04,
  node_size_factor = 5,
  color_by = "type",
  title = "Network Visualization",
  show_entity_types = TRUE,
  label_size = 1,
  verbose = TRUE
)

Arguments

Value

Invisible NULL (creates a file as a side effect)

Examples

# Create example results for visualization
results <- data.frame(
  a_term = rep("migraine", 4),
  b_term = c("serotonin", "CGRP", "cortisol", "dopamine"),
  c_term = c("sumatriptan", "fremanezumab", "propranolol", "amitriptyline"),
  abc_score = c(0.8, 0.7, 0.6, 0.5),
  b_type = c("chemical", "protein", "hormone", "chemical"),
  c_type = rep("drug", 4)
)


# These require graphics capabilities
plot_heatmap(results, output_file = tempfile(fileext = ".png"))
plot_network(results, output_file = tempfile(fileext = ".png"))

Description

This function ensures article data is valid for report generation, particularly handling publication years.

Usage

prep_articles(articles, verbose = TRUE)

Arguments

Value

A data frame of articles with validated publication years

Examples

# Create example article data
articles <- data.frame(
  title = c("Migraine Study 1", "Headache Research"),
  publication_year = c("2020", "not_a_year")
)

# Prepare articles
prepared <- prep_articles(articles)
print(prepared)

Description

This function preprocesses article text for further analysis.

Usage

preprocess_text(
  text_data,
  text_column = "abstract",
  remove_stopwords = TRUE,
  custom_stopwords = NULL,
  stem_words = FALSE,
  min_word_length = 3,
  max_word_length = 50
)

Arguments

Value

A data frame with processed text and extracted terms.

Examples

## Not run: 
processed_data <- preprocess_text(article_data, text_column = "abstract")

## End(Not run)

Description

This function searches PubMed using the NCBI E-utilities API via the rentrez package. The implementation includes optimizations for speed, memory efficiency, and reliability.

Usage

pubmed_search(
  query,
  max_results = 1000,
  use_mesh = FALSE,
  date_range = NULL,
  api_key = NULL,
  batch_size = 200,
  verbose = TRUE,
  use_cache = TRUE,
  retry_count = 3,
  retry_delay = 1
)

Arguments

Value

A data frame containing the search results with PubMed IDs, titles, and other metadata.

Examples

## Not run: 
results <- pubmed_search("migraine headache", max_results = 100)

## End(Not run)

Description

Query external biomedical APIs to validate entity types

Usage

query_external_api(term, claimed_type)

Arguments

Value

Logical indicating if the term was found in the appropriate database

Description

Query for MeSH terms using E-utilities

Usage

query_mesh(term, api_key = NULL)

Arguments

Value

A data frame with MeSH information for the term.

Description

Query UMLS for term information

Usage

query_umls(term, api_key, version = "current")

Arguments

Value

A data frame with UMLS information for the term.

Description

This function performs a comprehensive literature-based discovery analysis using multiple approaches without enforcing entity type constraints.

Usage

run_lbd(
  search_query,
  a_term,
  max_results = 100,
  discovery_approaches = c("abc", "anc", "lsi", "bitola"),
  include_visualizations = TRUE,
  output_file = "discovery_report.html",
  api_key = NULL,
  dictionary_sources = c("local", "mesh", "umls"),
  entity_categories = c("disease", "drug", "gene")
)

Arguments

Value

A list containing discovery results from all approaches.

Examples

## Not run: 
discovery_results <- run_lbd(
  search_query = "migraine headache",
  a_term = "migraine",
  discovery_approaches = c("abc", "anc", "lsi"),
  dictionary_sources = c("mesh"),
  entity_categories = c("disease", "drug", "gene", "protein")
)

## End(Not run)

Description

This function diversifies ABC results to avoid redundancy, with error handling to ensure results are always returned.

Usage

safe_diversify(
  top_results,
  diversity_method = "both",
  max_per_group = 5,
  min_score = 1e-04,
  min_results = 5,
  fallback_count = 15,
  verbose = TRUE
)

Arguments

Value

A data frame of diversified results

Examples

# Create example results
top_results <- data.frame(
  a_term = rep("migraine", 6),
  b_term = c("serotonin", "serotonin", "CGRP", "CGRP", "cortisol", "dopamine"),
  c_term = c("sumatriptan", "rizatriptan", "fremanezumab", "galcanezumab", "propranolol", "amitriptyline"),
  abc_score = c(0.8, 0.75, 0.7, 0.65, 0.6, 0.55)
)

# Apply diversification
diverse_results <- safe_diversify(top_results, max_per_group = 2)
print(diverse_results)

Description

This function sanitizes dictionary terms to ensure they're valid for entity extraction.

Usage

sanitize_dictionary(
  dictionary,
  term_column = "term",
  type_column = "type",
  validate_types = TRUE,
  verbose = TRUE
)

Arguments

Value

A data frame with sanitized terms.

Examples

# Create a dictionary with problematic terms
dirty_dict <- data.frame(
  term = c("migraine", "europe", "optimization", "receptor", "123", ""),
  type = c("disease", "location", "process", "protein", "number", "empty")
)

# Sanitize the dictionary
clean_dict <- sanitize_dictionary(dirty_dict)
print(clean_dict)

Description

This function saves search results to a file.

Usage

save_results(results, file_path, format = c("csv", "rds", "xlsx"))

Arguments

Value

The file path (invisibly).

Examples

## Not run: 
save_results(search_results, file_path = "search_results.csv")

## End(Not run)

Description

This function splits text into sentences.

Usage

segment_sentences(text)

Arguments

Value

A list where each element contains a character vector of sentences

Examples

## Not run: 
sentences <- segment_sentences(abstracts)

## End(Not run)

Description

This function applies validation to ensure only legitimate biomedical entities are included, while preserving trusted terms.

Usage

valid_entities(
  entities,
  primary_term,
  primary_term_variations = NULL,
  validation_function = NULL,
  verbose = TRUE,
  entity_col = "entity",
  type_col = "entity_type"
)

Arguments

Value

A data frame of filtered entities

Examples

# Create example entities
entities <- data.frame(
  entity = c("migraine", "optimization", "receptor", "europe"),
  entity_type = c("disease", "process", "protein", "location")
)

# Validate entities
validated <- valid_entities(entities, "migraine", c("migrain", "headache"))
print(validated)

Description

This function performs statistical tests to validate ABC model results. It calculates p-values using hypergeometric tests and applies correction for multiple testing. The function is optimized to work with very large co-occurrence matrices.

Usage

validate_abc(
  abc_results,
  co_matrix,
  alpha = 0.05,
  correction = c("BH", "bonferroni", "none"),
  filter_by_significance = FALSE
)

Arguments

Value

A data frame with ABC results and statistical significance measures.

Examples

## Not run: 
validated_results <- validate_abc(abc_results, co_matrix)

## End(Not run)

Description

Validate biomedical entities using BioBERT or other ML models

Usage

validate_biomedical_entity(term, claimed_type)

Arguments

Value

Logical indicating if the term is validated

Description

Comprehensive entity validation using multiple techniques

Usage

validate_entity_comprehensive(
  term,
  claimed_type,
  use_nlp = TRUE,
  use_pattern = TRUE,
  use_external_api = FALSE
)

Arguments

Value

Logical indicating if the term is validated

Description

Validate entity types using NLP-based entity recognition with improved accuracy

Usage

validate_entity_with_nlp(term, claimed_type, nlp_model = NULL)

Arguments

Value

Logical indicating if the term is likely of the claimed type

Description

This function validates a UMLS API key using the validation endpoint.

Usage

validate_umls_key(api_key, validator_api_key = NULL)

Arguments

Value

Logical indicating if the API key is valid

Examples

## Not run: 
is_valid <- validate_umls_key("user_api_key")

## End(Not run)

Description

This function preprocesses text data using vectorized operations for better performance.

This function preprocesses text data using vectorized operations for better performance.

Usage

vec_preprocess(
  text_data,
  text_column = "abstract",
  remove_stopwords = TRUE,
  custom_stopwords = NULL,
  min_word_length = 3,
  max_word_length = 50,
  chunk_size = 100
)

vec_preprocess(
  text_data,
  text_column = "abstract",
  remove_stopwords = TRUE,
  custom_stopwords = NULL,
  min_word_length = 3,
  max_word_length = 50,
  chunk_size = 100
)

Arguments

Value

A data frame with processed text.

A data frame with processed text.

Examples

## Not run: 
processed_data <- vec_preprocess(article_data, text_column = "abstract")

## End(Not run)
## Not run: 
processed_data <- vec_preprocess(article_data, text_column = "abstract")

## End(Not run)

Description

This function creates a heatmap visualization of ABC connections using base R graphics.

Usage

vis_abc_heatmap(
  abc_results,
  top_n = 25,
  min_score = 0.1,
  show_labels = TRUE,
  title = "ABC Connections Heatmap"
)

Arguments

Value

NULL invisibly. The function creates a plot as a side effect.

Examples

## Not run: 
vis_abc_heatmap(abc_results, top_n = 20)

## End(Not run)

Description

This function creates an improved heatmap visualization of ABC connections that can display entity type information when available, without enforcing type constraints.

Usage

vis_heatmap(
  abc_results,
  top_n = 25,
  min_score = 0.1,
  show_significance = TRUE,
  color_palette = "blues",
  title = "ABC Connections Heatmap",
  show_entity_types = TRUE
)

Arguments

Value

NULL invisibly. The function creates a plot as a side effect.

Examples

## Not run: 
vis_heatmap(abc_results, top_n = 20, show_significance = TRUE)

## End(Not run)

Description

This function creates an improved network visualization of ABC connections that displays entity types when available, without enforcing type constraints.

Usage

vis_network(
  abc_results,
  top_n = 25,
  min_score = 0.1,
  show_significance = TRUE,
  node_size_factor = 5,
  color_by = "type",
  title = "ABC Model Network",
  show_entity_types = TRUE,
  label_size = 1
)

Arguments

Value

NULL invisibly. The function creates a plot as a side effect.

Examples

## Not run: 
vis_network(abc_results, top_n = 20, show_significance = TRUE,
                     show_entity_types = TRUE)

## End(Not run)

Description

Create a network visualization of ABC connections using base R graphics.

Usage

vis_abc_network(
  abc_results,
  top_n = 25,
  min_score = 0.1,
  node_size_factor = 3,
  edge_width_factor = 1,
  color_by = "type",
  title = "ABC Model Network"
)

Arguments

Value

NULL invisibly. The function creates a plot as a side effect.

Examples

## Not run: 
# Create a network visualization of ABC model results
vis_abc_network(abc_results, top_n = 20)

## End(Not run)