public async Task Index()
{
var reader = await FastaReader.CreateAsync(FastaPath);
foreach (var identifier in reader.Identifiers)
{
var comment = reader.CommentOf(identifier);
var attributes = AttributesParser.Parse(comment);
var id = attributes.GetValueOrDefault("protein_id");
if (id == null)
{
throw new Exception("Id missing");
}
var range = ParseRange(attributes.GetValueOrDefault("location", ""));
var data = new GeneData(id, SequenceId)
{
Symbol = attributes.GetValueOrDefault("gene"),
Product = attributes.GetValueOrDefault("protein"),
LocusTag = attributes.GetValueOrDefault("locus_tag") ?? throw new Exception("Locus tag missing"),
// We can infer this from the fasta reader if we prevent '\n' from being in the sequence
Position = range?.Item1,
Length = range != null ? range?.Item2 - range?.Item1 + 1 : null
};
await Database.Index(data);
}
}
}