public IndexedDatabase(FastaDatabase fastaDatabase)
{
FastaDatabase = fastaDatabase;
var databaseFilePath = FastaDatabase.GetFastaFilePath();
//var databaseFilePathNoExt = Path.Combine(Path.GetDirectoryName(databaseFilePath), Path.GetFileNameWithoutExtension(databaseFilePath));
//_pLcpFilePath = databaseFilePathNoExt + PermutedLongestCommonPrefixFileExtension;
_pLcpFilePath = Path.ChangeExtension(databaseFilePath, PermutedLongestCommonPrefixFileExtension);
var lastWriteTimeHash = FastaDatabase.GetLastWriteTimeHash();
if (!File.Exists(_pLcpFilePath) || !FastaDatabase.CheckHashCodeBinaryFile(_pLcpFilePath, lastWriteTimeHash))
{
Console.Write("Generating " + _pLcpFilePath + " ... ");
CreatePermutedLongestCommonPrefixFile();
Console.WriteLine("Done");
}
}
/// <summary>
/// Constructor - build the index
/// </summary>
/// <param name="fastaDatabase"></param>
public IndexedDatabase(FastaDatabase fastaDatabase)
{
FastaDatabase = fastaDatabase;
var databaseFilePath = FastaDatabase.GetFastaFilePath();
//var databaseFilePathNoExt = Path.Combine(Path.GetDirectoryName(databaseFilePath), Path.GetFileNameWithoutExtension(databaseFilePath));
//_pLcpFilePath = databaseFilePathNoExt + PermutedLongestCommonPrefixFileExtension;
_pLcpFilePath = Path.ChangeExtension(databaseFilePath, PermutedLongestCommonPrefixFileExtension);
var lastWriteTimeHash = FastaDatabase.GetLastWriteTimeHash();
if (!File.Exists(_pLcpFilePath) || !FastaDatabase.CheckHashCodeBinaryFile(_pLcpFilePath, lastWriteTimeHash))
{
Console.Write("Generating " + _pLcpFilePath + " ... ");
CreatePermutedLongestCommonPrefixFile();
Console.WriteLine("Done");
}
}
private void CreatePermutedLongestCommonPrefixFile()
{
if (File.Exists(_pLcpFilePath))
{
File.Delete(_pLcpFilePath);
}
var sequence = FastaDatabase.GetSequence();
//Console.WriteLine("Annotation: {0}", System.Text.Encoding.ASCII.GetString(sequence));
var suffixArray = new int[sequence.Length - 1];
SAIS.sufsort(sequence, suffixArray, sequence.Length - 1);
var prevIndex = sequence.Length - 1;
var pLcp = new byte[suffixArray.Length];
// Data dependency: cannot run in parallel
foreach (var index in suffixArray)
{
var lcp = GetLcp(sequence, prevIndex, index);
pLcp[index] = lcp;
//Console.WriteLine("{0}\t{1}", System.Text.Encoding.ASCII.GetString(sequence, offset, sequence.Length - offset - 1), lcp);
prevIndex = index;
}
using (var fs = new FileStream(_pLcpFilePath, FileMode.OpenOrCreate, FileAccess.Write))
{
foreach (var lcp in pLcp)
{
//Console.WriteLine("LCP: {0}", lcp);
fs.WriteByte(lcp);
}
fs.Write(BitConverter.GetBytes(FastaDatabase.FileFormatId), 0, sizeof(int));
fs.Write(BitConverter.GetBytes(FastaDatabase.GetLastWriteTimeHash()), 0, sizeof(int));
}
}