DAVID (Database for Annotation, Visualization, and Integrated Discovery)

A core algorithm that groups redundant, highly similar annotation terms into clusters to reduce the complexity of enrichment results. It measures the kappa statistic between gene-term associations to quantify co-occurrence, then uses a heuristic fuzzy clustering approach to group related terms. This prevents the output from being dominated by multiple overlapping Gene Ontology terms describing the same biological process, providing a concise, high-level view of the underlying biology.

A traditional enrichment view that reports the statistical significance of individual annotation terms associated with a gene list. It employs a modified Fisher's Exact Test (EASE Score) to calculate p-values, providing a conservative adjustment to standard over-representation analysis. Results are sortable by p-value, fold enrichment, and false discovery rate, allowing rapid identification of the most significantly enriched biological themes.

A gene-centric view that displays all annotation terms associated with each gene in the input list. This module allows users to inspect the specific functional roles of individual genes within the context of the enrichment results. It links directly to the original database sources, enabling deep dives into the evidence supporting each gene-term association.

A translation module that resolves the identifier mapping problem by converting between diverse gene and protein identifier systems. It supports dozens of formats including Entrez Gene ID, UniProt Accession, Ensembl Gene ID, and official gene symbols. This is a critical preprocessing step, as enrichment analysis requires a unified identifier namespace for accurate background population calculations.

DAVID aggregates annotation data from a curated set of authoritative sources into a unified knowledgebase. Key integrated databases include:

• Gene Ontology (GO) for biological process, molecular function, and cellular component
• KEGG and Reactome for pathway mapping
• InterPro for protein domains and families
• Disease associations from OMIM and other sources This multi-source integration provides a comprehensive functional landscape from a single query.

A module that groups genes based on the similarity of their functional annotation profiles rather than grouping annotation terms. It uses a co-occurrence matrix to measure functional relatedness between genes, then applies a heuristic clustering algorithm to identify functionally coherent gene modules. This is particularly useful for identifying novel members of a biological pathway or protein complex within a list of differentially expressed genes.

The foundational statistical engine underlying DAVID's core workflow. ORA uses a hypergeometric distribution or Fisher's exact test to determine if a user's gene list contains more members of a specific annotation category than would be expected by random chance. DAVID automates this process across thousands of GO terms, KEGG pathways, and protein domains simultaneously, applying multiple hypothesis testing correction to control false positives.

A species-independent controlled vocabulary that DAVID relies on as its primary annotation backbone. GO is structured into three orthogonal ontologies:

• Biological Process: Pathways and larger cellular programs
• Molecular Function: Catalytic and binding activities at the protein level
• Cellular Component: Subcellular locations and macromolecular complexes DAVID maps user gene lists to GO terms and identifies statistically enriched categories, providing the functional interpretation layer.

DAVID's proprietary algorithm that distinguishes it from simpler enrichment tools. Rather than returning a flat, redundant list of enriched terms, DAVID groups related annotations into clusters based on gene co-occurrence. This reduces the cognitive load of interpreting hundreds of overlapping GO terms by presenting a condensed, biologically meaningful summary. Each cluster receives an enrichment score reflecting its overall significance.

The primary tabular output in DAVID presenting a ranked list of enriched annotation terms. Each row includes:

• Term name and source database (GO, KEGG, INTERPRO, etc.)
• Gene count and percentage of user list mapping to the term
• P-value from the modified Fisher's exact test
• Benjamini-Hochberg adjusted p-value for FDR control This structured report enables rapid identification of the most significantly affected biological themes.

A complementary methodology to DAVID's ORA approach. While DAVID requires a pre-filtered list of differentially expressed genes, GSEA operates on a ranked list of all genes without arbitrary cutoffs. GSEA detects subtle, coordinated changes in pathway activity that ORA might miss when individual genes fall below significance thresholds. Many researchers use both tools sequentially for comprehensive functional interpretation.