public async Task Index()
{
var list = await GffFileReader.FromFile(GffPath);
foreach (var gff in list)
{
var attributes = gff.Attributes;
var id = attributes.GetValueOrDefault("protein_id");
if (id == null)
{
continue;
}
var note = attributes.GetValueOrDefault("Note");
var product = attributes.GetValueOrDefault("product");
var data = new GeneData(id, gff.Sequence)
{
Symbol = attributes.GetValueOrDefault("gene"),
Product = product != null?Uri.UnescapeDataString(product) : null,
LocusTag = attributes.GetValueOrDefault("locus_tag") ?? throw new Exception("Locus tag missing"),
Position = gff.Start,
Length = gff.End - gff.Start + 1,
Strand = gff.Strand,
Score = gff.Score,
Phase = gff.Phase,
Note = note != null?Uri.UnescapeDataString(note) : null
};
await Database.Index(data);
}
}
}
public async Task IndexSnp(SnpSource source, Snp snp)
{
if (Genes == null)
{
Genes = await Database.GetAllGene();
}
var doi = source.Study?.Doi;
SnpData snpData = SnpData.FromSnp(snp);
if (doi != null)
{
var citation = await DoiClient.GetCitation(doi);
if (citation != null)
{
snpData.Annotations.Add(new SnpData.Annotation
{
Study = citation,
Lineage = snp.Attributes.GetValueOrDefault("lineage")
});
}
}
;
snpData = snpData with
{
Gene = Genes
.FirstOrDefault(v => v.Position <snp.Position && v.Position + v.Length> snp.Position)
};
await Database.Index(snpData);
}
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);
}
}
}
