Gene Ontology (GO) evidence codes

The Gene Ontology (GO) evidence codes describe the source and reliability of GO annotations. Each code represents a different evidence source or research method, helping researchers understand the reliability and basis of the annotation. Evidence codes are categorized into six main types:
1. Experimentally Supported Evidence
EXP (Inferred from Experiment): Experimental validation evidence, indicating that the annotation comes from direct experimental data.
IDA (Inferred from Direct Assay): Direct assay evidence, typically obtained through biochemical or other direct methods (e.g., activity assays, binding assays).
IPI (Inferred from Physical Interaction): Physical interaction evidence based on direct interactions between proteins, such as co-immunoprecipitation (Co-IP) or yeast two-hybrid.
IMP (Inferred from Mutant Phenotype): Evidence from mutant phenotype, based on phenotypic changes observed in gene knockout or mutation experiments. For instance, deleting a gene resulting in abnormal cell growth suggests it may influence growth.
IGI (Inferred from Genetic Interaction): Genetic interaction evidence, based on interactions between multiple genes, such as observing a stronger phenotype in double mutants than single mutants.
IEP (Inferred from Expression Pattern): Evidence from expression patterns, based on gene or protein expression changes under specific conditions or times, often detected by techniques like Northern blot or RNA-seq.
2. High-Throughput Experiment Supported Evidence
HTP (High Throughput Experiment): General large-scale experiment, indicating data from high-throughput methods like genome sequencing.
HDA (High Throughput Direct Assay): High-throughput direct assay evidence, from large-scale direct experiments.
HMP (High Throughput Mutant Phenotype): High-throughput mutant phenotype evidence, from large-scale mutant phenotype analysis.
HGI (High Throughput Genetic Interaction): High-throughput genetic interaction data, e.g., gene interaction networks from large-scale screenings.
HEP (High Throughput Expression Pattern): High-throughput expression pattern evidence, based on large-scale gene expression experiments (e.g., RNA-seq) for analyzing gene expression in different conditions or tissues.
3. Computational Analysis Supported Evidence
ISS (Inferred from Sequence or Structural Similarity): Sequence or structural similarity evidence, based on the similarity between the target gene and known functional genes, typically obtained through bioinformatics tools (e.g., BLAST).
ISO (Inferred from Sequence Orthology): Sequence orthology evidence, based on homology among genes within the same species, often for conserved gene functions.
ISA (Inferred from Sequence Alignment): Sequence alignment evidence, based on sequence alignment, usually for genes with similar domains in different species.
ISM (Inferred from Sequence Model): Sequence model evidence, inferred from recognizing specific sequence motifs or patterns.
IGC (Inferred from Genomic Context): Genomic context evidence, based on gene location similarity across species to infer function.
RCA (Inferred from Reviewed Computational Analysis): Reviewed computational analysis evidence, based on manually reviewed computational analysis results.
4. Author Statements as Evidence
TAS (Traceable Author Statement): Traceable author statement, based on statements in published literature with specific data support.
NAS (Non-traceable Author Statement): Non-traceable author statement, based on published statements but without specific data support, generally considered less reliable.
5. Curatorial and Inferred Evidence
IC (Inferred by Curator): Curator-inferred evidence, conclusions made by database curators based on logical reasoning combined with other evidence sources.
ND (No biological Data available): No biological data available; currently, there is no supporting experimental or computational data for the annotation.
6. Electronic Annotation Evidence
IEA (Inferred from Electronic Annotation): Electronic annotation evidence, generated by automated computational methods without manual review, often used in high-throughput preliminary annotation.