SO property for GFF "source" field

[arnikz]

Hi, I'd like to link the "source" field of the GFF (2nd column) to Sequence Ontology but can't find a suitable property to do that (e.g. has_origin has a different purpose). For example, DDBJ/ENA/GenBank Feature Table defines an optional qualifier "inference", which might refer to a gene prediction tool etc. Any ideas on this? Thanks.

[keilbeck]

I'm not quite sure I understand. You want to link the source to the type. Would it always be the same or would there be multiples types of linkage? infers, predicts, states etc?

[arnikz]

In my GFF file there are, for example, gene features predicted by the Eugene software (indicated in the source field). Similarly to the gene feature type (SO_0000704). I would like to link the source to some SO type. I think, suitable terms could be is output of or is inferred_by. In case the features in the GFF file originate directly from a database (e.g. UniProt or GenBank) one could use, e.g. imported_from or originates_from terms. What do you think?

[arnikz]

Hi @keilbeck, any suggestions wrt. SO and source field in GFF?

[nicoleruiz]

One possible solution would be to create a generic GFF source term and then have different relationships that capture if the sequence_feature was predicted by a specific tool or if it came from a specific database. For example, you could have sequence_feature is_predicted_by gff_source or sequence_feature imported_from gff_source.