private static string ApplyRegex(FastaHeaderFieldRegex regex, string line)
{
string result = null;
if (regex != null)
{
var matches = regex.Regex.Matches(line);
if (matches.Count > regex.Match && matches[regex.Match].Groups.Count > regex.Group)
{
result = matches[regex.Match].Groups[regex.Group].Value;
}
}
return(result);
}
public void Test_accession_regex_weird()
{
FastaHeaderFieldRegex bad = new FastaHeaderFieldRegex("", @"/()/", 0, 1);
List <Protein> ok = ProteinDbLoader.LoadProteinFasta(Path.Combine(TestContext.CurrentContext.TestDirectory, "DatabaseTests", @"test_ensembl.pep.all.fasta"), true, DecoyType.None, false,
bad, bad, bad, bad, bad, out var a);
ProteinDbWriter.WriteFastaDatabase(ok, Path.Combine(TestContext.CurrentContext.TestDirectory, "DatabaseTests", @"rewrite_test_ensembl.pep.all.fasta"), " ");
List <Protein> ok2 = ProteinDbLoader.LoadProteinFasta(Path.Combine(TestContext.CurrentContext.TestDirectory, "DatabaseTests", @"rewrite_test_ensembl.pep.all.fasta"), true, DecoyType.None, false,
bad, bad, bad, bad, bad, out var b);
Assert.AreEqual("ENSP00000381386 pep:known chromosome:GRCh37:22:24313554:24316773:-1 gene:ENSG00000099977 transcript:ENST00000398344 gene_biotype:protein_coding transcript_biotype:protein_coding", ok[0].Accession);
Assert.AreEqual("ENSP00000381386 pep:known chromosome:GRCh37:22:24313554:24316773:-1 gene:ENSG00000099977 transcript:ENST00000398344 gene_biotype:protein_coding transcript_biotype:protein_coding", ok2[0].Accession);
Assert.AreEqual(ok.Count, ok2.Count);
Assert.True(Enumerable.Range(0, ok.Count).All(i => ok[i].BaseSequence == ok2[i].BaseSequence));
}
