private static void ProcessPage(SNPDatabase database, string pageTitle, string pageText)
{
if (pageText != null)
{
// Try to parse the page's SNP category properties.
var snpInfo = ParseSNPPage(pageText);
if (snpInfo != null)
{
// Skip pages with mismatched SNP ID and title.
if (snpInfo.Value.id == pageTitle.ToLowerInvariant())
{
database.snpToInfoMap.Add(snpInfo.Value.id, snpInfo.Value);
}
}
else
{
var genotypeCategoryText = ParsePageCategorySubstring(pageText, "Genotype");
if (genotypeCategoryText != null)
{
// Parse the trait summary for a SNP genotype and add it to the previously created SNPInfo.
var rsid = string.Format("rs{0}", ParseCategoryProperty(genotypeCategoryText, "rsid"));
var genotype = DNA.StringToDiploidGenotype(string.Format("{0};{1}", ParseCategoryProperty(genotypeCategoryText, "allele1"), ParseCategoryProperty(genotypeCategoryText, "allele2")));
var genotypeTraitSummary = ParseCategoryProperty(genotypeCategoryText, "summary");
if (database.snpToInfoMap.ContainsKey(rsid))
{
var snpGenotypes = database.snpToInfoMap[rsid].genotypes;
for (var genotypeIndex = 0; genotypeIndex < snpGenotypes.Length; ++genotypeIndex)
{
if (snpGenotypes[genotypeIndex].genotype.Equals(genotype))
{
snpGenotypes[genotypeIndex].trait = genotypeTraitSummary;
break;
}
}
}
}
else if (ParsePageCategorySubstring(pageText, "is a \\| medical condition") != null ||
ParsePageCategorySubstring(pageText, "is a \\| medicine") != null ||
ParsePageCategorySubstring(pageText, "is a \\| gene") != null)
{
// If the page isn't a SNP, add it to the database as a trait with associated SNPs.
var traitInfo = new TraitInfo();
traitInfo.title = pageTitle;
// Parse links to associated SNPs.
var associatedSNPs = new List <string>();
Utilities.ProcessDelimitedItems(pageText, "[[", "]]", delegate(string itemText)
{
var snpId = itemText.ToLowerInvariant();
if (snpId.StartsWith("rs"))
{
if (!associatedSNPs.Contains(snpId))
{
associatedSNPs.Add(snpId);
}
}
return(itemText);
});
traitInfo.associatedSNPs = associatedSNPs.ToArray();
if (traitInfo.associatedSNPs.Length > 0)
{
// Add the trait to the database.
database.traits.Add(traitInfo);
}
}
}
}
}
