Gff Structure

Key attributes

• ID — always prefixed with gene: (e.g., gene:ENSSSAG00000110109)
• Name — optional, only ~44% of genes have it
• biotype — e.g., protein_coding, lncRNA, pseudogene, snRNA, snoRNA, miRNA, tRNA, rRNA
• description — URL-encoded, contains source in bracket suffix: [Source:ZFIN%3BAcc:ZDB-GENE-030131-551]
• gene_id — same as ID without gene: prefix
• logic_name — ensembl or ncrna

Description sources (extracted from [Source:XXX;Acc:YYY])

• ZFIN — zebrafish nomenclature (most common)
• RFAM — RNA families
• NCBI gene
• HGNC — human gene nomenclature
• miRBase

Naming patterns

• If a gene has Name=, it always also has description=
• No genes have description= without Name=
• lncRNA genes are essentially never named

NCBI GFF3

Gene-level feature type

• Only gene (no ncRNA_gene/pseudogene distinction at feature level)

Attributes format (col 9)

ID=gene-LOC123729278;Dbxref=GeneID:123729278;Name=LOC123729278;description=lethal(3)malignant brain tumor-like protein 2;gbkey=Gene;gene=LOC123729278;gene_biotype=protein_coding

Key attributes

• ID — prefixed with gene- (e.g., gene-LOC123729278 or gene-wdr32)
• Name — always present, but often auto-assigned LOC IDs
• Dbxref — contains GeneID:NNNNN (NCBI numeric gene ID)
• gene_biotype — (note: NOT biotype like Ensembl) e.g., protein_coding, lncRNA, tRNA, rRNA, snRNA, snoRNA, misc_RNA
• description — plain text, URL-encoded (e.g., %2C for comma)
• gene — same as Name
• gbkey — always Gene

Naming patterns

• ALL genes have Name= but most are auto-assigned LOC IDs (start with "LOC")
• Real gene symbols are lowercase (e.g., wdr32, manba)
• LOC genes: description may be informative or "uncharacterized LOCnnn"

Distinguishing real names from LOC IDs

• Check if Name starts with "LOC" followed by digits
• is_loc = name.startswith("LOC")

Key Differences Between Ensembl and NCBI

Seqid Values

• Both use chromosome numbers: 1, 2, ..., 29
• Plus unplaced scaffolds

NCBI Gene Orthologs (ftp.ncbi.nlm.nih.gov/gene/DATA/gene_orthologs.gz)

• Pairs stored directionally: "reference" organism in tax_id, "other" in Other_tax_id
• Atlantic salmon (tax_id=8030) is always in Other_tax_id column (never in tax_id)
• Human-salmon pairs: filter tax_id == 9606 & Other_tax_id == 8030
• GeneID = human gene, Other_GeneID = salmon gene
• Header starts with #tax_id — needs renaming after read