internal Protein(ProteomeDbPath proteomeDb, DbProtein protein, DbProteinName primaryName)
: base(proteomeDb, protein)
{
Sequence = protein.Sequence;
if (primaryName != null)
{
_proteinMetadata = primaryName.GetProteinMetadata();
if (primaryName.Protein != null)
{
// grab the alternative names now, rather than going back to the db later
_alternativeNames = new List <ProteinMetadata>();
foreach (var name in primaryName.Protein.Names)
{
if (!name.IsPrimary)
{
_alternativeNames.Add(name.GetProteinMetadata());
}
}
}
}
}
/// <summary>
/// Uses the known list of regexes to parse lineIn, keeping the
/// result that fills in the most metdata. In the event of a tie,
/// first result wins - so regex list order matters.
/// Populates the WebSearchInfo field but does not perform
/// the actual search - that's done elsewhere.
/// </summary>
/// <param name="lineIn">the text to be parsed</param>
public ProteinMetadata ParseProteinMetaData(String lineIn)
{
if (lineIn.Length <= 0)
return null;
var line = lineIn.Replace('\t', ' '); // regularize whitespace for simpler regexes
// If there is a second >, then this is a custom name, and not
// a real FASTA sequence.
int start = (line.Length > 0 && line[0] == '>' ? 1 : 0);
if (line.Length > 1 && line[1] == '>')
{
start++;
}
ProteinMetadata bestResult = null;
var bestCount = 0;
foreach (var r in _regexFasta)
{
Match match = r.RegexPattern.Match(line.Substring(start));
if (match.Success)
{
// a hit - now use the replacement expression to get the ProteinMetadata parts
string[] regexOutputs = r.RegexPattern.Replace(line.Substring(start), r.RegexReplacement).Split('\n');
var headerResult = new DbProteinName();
string searchterm = null; // assume no webservice lookup unless told otherwise
int dbColumnsFound = 0;
for (var n = regexOutputs.Length; n-- > 0;)
{
var split = regexOutputs[n].Split(new[] {':'}, 2); // split on first colon only
if (split.Length == 2)
{
var type = split[0].Trim();
var val = split[1].Trim();
if (val.Contains("${")) // failed match // Not L10N
{
val = String.Empty;
}
if (val.Length > 0)
{
dbColumnsFound++; // valid entry
switch (type)
{
case "name": // Not L10N
headerResult.Name = val;
break;
case "description": // Not L10N
headerResult.Description = val;
break;
case "accession": // Not L10N
headerResult.Accession = val;
break;
case "preferredname": // Not L10N
headerResult.PreferredName = val;
break;
case "gene": // Not L10N
headerResult.Gene = val;
break;
case "species": // Not L10N
headerResult.Species = val;
break;
case "searchterm": // Not L10N
dbColumnsFound--; // not actually a db column
searchterm = val;
break;
default:
throw new ArgumentOutOfRangeException(
String.Format("Unknown Fasta RegEx output formatter type \'{0}\'", // Not L10N
regexOutputs[n]));
}
}
}
else
{
throw new ArgumentOutOfRangeException(
String.Format("Fasta RegEx failure in \'{0}\'", // Not L10N
line.Substring(start)));
}
}
if (headerResult.GetProteinMetadata().HasMissingMetadata())
{
if (searchterm != null)
{
// shave off any alternatives (might look like "IPI:IPI00197700.1|SWISS-PROT:P04638|ENSEMBL:ENSRNOP00000004662|REFSEQ:NP_037244")
searchterm = searchterm.Split('|')[0];
// a reasonable accession value will have at least one digit in it, and won't have things like tabs and parens and braces that confuse web services
if ("0123456789".Any(searchterm.Contains) && !" \t()[]".Any(searchterm.Contains)) // Not L10N
headerResult.SetWebSearchTerm(new WebSearchTerm(searchterm[0], searchterm.Substring(1))); // we'll need to hit the webservices to get this missing info
}
}
if (headerResult.GetProteinMetadata().WebSearchInfo.IsEmpty())
headerResult.SetWebSearchCompleted(); // no search possible
if (dbColumnsFound > bestCount)
{
bestCount = dbColumnsFound; // best match so far - tie goes to the first hit so order matters
bestResult = headerResult.GetProteinMetadata();
}
}
}
return bestResult;
}
