is_allele_of gene allele A genomic feature that represents one of a set of versions of a gene (i.e. a haplotype whose extent is that of a gene) In SO, the concept of a 'gene' is functionally defined, in that a gene necessarily produces a functional product. By contrast, the concept of a 'gene allele' here is positionally defined - representing the sequence present at the location a gene resides in a reference genome (based on sequence alignment). An Shh gene allele, for example, may be a fully functional wild-type version of the gene, a non-functional version carrying a deleterious point mutation, a truncated version of the gene, or even a complete deletion. In all these cases, an 'Shh gene allele' exists at the position where the canonical gene resides in the reference genome - even if the extent of this allele different than the wild-type, or even zero in the case of the complete deletion. A genomic feature being an allele_of a gene is based on its location in a host genome - not on its sequence. This means, for example, that the insertion of the human SMN2 gene into the genome of a mouse (see http://www.informatics.jax.org/allele/MGI:3056903) DOES NOT represent an allele_of the human SMN2 gene according to the GENO model - because it is located in a mouse genome, not a human one. Rather, this is a transgenic insertion that derives_sequence_from the human SMN2 gene. If this human SMN2 gene is inserted within the mouse SMN2 gene locus (e.g. used to replace mouse SMN2 gene), the feature it creates is an allele_of the mouse SMN2 gene (one that happens to match the sequence of the human ortholog of the gene). But again, it is not an allele_of the human SMN2 gene. Regarding the distinction between a 'gene' and a 'gene allele': Every zebrafish genome contains a 'gene allele' for every zebrafish gene. Many will be 'wild-type' or at least functional gene alleles. But some may be alleles that are mutated or truncated so as to lack functionality. According to current SO criteria defining genes, a 'gene' no longer exists in the case of a non-functional or deleted variant. But the 'gene allele' does exist - and its extent is that of the remaining/altered sequence based on alignment with a reference gene. Even for completely deleted genes, an allele of the gene exists (and here is equivalent to the junction corresponding to the where gene would live based on a reference alignment). This design allows us to classify genes and any variants of those genes (be they functional or not) as the same type of thing (ie a 'gene allele'), since classification is based on genomic position rather than functional capacity. This is practical for representation of variant genotypes which often carry non-functional versions of a gene at a particular locus. What is important here is specifying what is present at a locus associated with a particular gene, whether or not it is a functional gene or not. http://purl.obolibrary.org/obo/SO_0001023 ! allele The reference/wild-type cd99l2 danio rerio gene allele spans bases 27,004,426-27,021,059 on Chromosome 7. The "mn004Gt" represents an experimentally-created allele of this gene, in which sequence from a gene trap construct containing an RFP marker has been inserted at the cd99l2 gene locus. The resulting gene allele includes sequence from this construct that make it longer than the reference gene sequence, and also alter its seqauence in a way that prevents it from producing a functional product. The sequence extent of this cd99l2 gene allele is determined based on how its sequence aligns with that of the canonical gene and surrounding sequence in a reference genome. http://useast.ensembl.org/Danio_rerio/Gene/Summary?g=ENSDARG00000056722 https://zfin.org/ZDB-ALT-111117-8 allele gene