CSMSL.Examples.ProteinGroupingExample.ExampleProteinGrouping C# (CSharp) Method

        public static void ExampleProteinGrouping(IProtease protease, double percentIdentified = 0.01, int maxMissed = 3, int minLength = 5, int maxLength = 50)
        {
            Stopwatch watch = new Stopwatch();
            watch.Start();
            List<Peptide> peps = new List<Peptide>(1000000);
            List<Protein> proteins = new List<Protein>(7000);

            using (FastaReader reader = new FastaReader("Resources/yeast_uniprot_120226.fasta"))
            {
                foreach (Protein protein in reader.ReadNextProtein())
                {
                    peps.AddRange(protein.Digest(protease, maxMissed, minLength, maxLength));
                    proteins.Add(protein);
                }
            }
            Console.WriteLine("Loaded {0:N0} peptides from {1:N0} proteins in {2} ms", peps.Count, proteins.Count, watch.ElapsedMilliseconds);
            watch.Restart();

            Random random = new Random(480912341);

            // Take the first x % to act as our identified peptides
            List<Peptide> identifiedPeptides = peps.OrderBy(x => random.Next()).Take((int) (peps.Count*percentIdentified)).ToList();

            List<ProteinGroup> proteinGroups = ProteinGroup.GroupProteins(proteins, protease, identifiedPeptides, new AminoAcidLeucineSequenceComparer(), maxMissed).ToList();

            watch.Stop();

            Console.WriteLine("{0:N0} proteins produced {1:N0} protein groups from {2:N0} identified sequences", proteins.Count, proteinGroups.Count, identifiedPeptides.Count);
            Console.WriteLine();
            Console.WriteLine("Time elapsed: {0} ms", watch.ElapsedMilliseconds);
        }