fixe filter file reading

src/main/java/edu/umd/marbl/mhap/main/MhapMain.java

@@ -57,7 +57,7 @@ public final class MhapMain
 {
 	private final double acceptScore;
 
-	private final HashSet<Integer> filter;
+	private final HashSet<Long> filter;
 
 	private final String inFile;
 
@@ -289,9 +289,6 @@ public MhapMain(ParseOptions options) throws IOException
 		this.minStoreLength = options.get("--min-store-length").getInteger();
 		this.maxShift = options.get("--max-shift").getDouble();
 		this.acceptScore = options.get("--threshold").getDouble();
-
-		this.kmerCounter = recordFastaKmerCounts(inFile, options.get("--filter-threshold").getDouble());
-		//this.kmerCounter = null;
 
 		// read in the kmer filter set
 		String filterFile = options.get("-f").getString();
@@ -309,9 +306,14 @@ public MhapMain(ParseOptions options) throws IOException
 				throw new MhapRuntimeException("Could not parse k-mer filter file.", e);
 			}
 			System.err.println("Time (s) to read filter file: " + (System.nanoTime() - startTime) * 1.0e-9);
+
+			this.kmerCounter = null;
 		}
 		else
+		{
 			this.filter = null;
+			this.kmerCounter = recordFastaKmerCounts(inFile, options.get("--filter-threshold").getDouble());
+		}
 
 	}
 

src/main/java/edu/umd/marbl/mhap/sketch/MinHash.java

@@ -58,7 +58,7 @@ public final class MinHash implements Serializable
 
 
 	public final static int[] computeKmerMinHashesWeightedIntSuper(String seq, final int kmerSize, final int numHashes,
-			HashSet<Integer> filter, KmerCounts kmerCount)
+			HashSet<Long> filter, KmerCounts kmerCount)
 	{
 		final int numberKmers = seq.length() - kmerSize + 1;
 
@@ -241,7 +241,7 @@ private MinHash(int seqLength, int[] minHashes)
 		this.minHashes = minHashes;
 	}
 
-	public MinHash(Sequence seq, int kmerSize, int numHashes, HashSet<Integer> filter, KmerCounts kmerCount)
+	public MinHash(Sequence seq, int kmerSize, int numHashes, HashSet<Long> filter, KmerCounts kmerCount)
 	{
 		this.seqLength = seq.length();
 

src/main/java/edu/umd/marbl/mhap/sketch/MinHashSearch.java

@@ -163,7 +163,7 @@ public boolean addSequence(SequenceSketch currHash)
 	public List<MatchResult> findMatches(SequenceSketch seqHashes, boolean toSelf)
 	{
 		//for performance reasons might need to change
-		//long startTime = System.nanoTime();
+		long startTime = System.nanoTime();
 
 		MinHash minHash = seqHashes.getMinHashes();
 
@@ -204,8 +204,8 @@ public List<MatchResult> findMatches(SequenceSketch seqHashes, boolean toSelf)
 		}
 
 		//record the search time
-		//long minHashEndTime = System.nanoTime();
-		//this.minhashSearchTime.getAndAdd(minHashEndTime - startTime);
+		long minHashEndTime = System.nanoTime();
+		this.minhashSearchTime.getAndAdd(minHashEndTime - startTime);
 
 		//record number of hash matches processed
 		this.numberSequencesHit.getAndAdd(bestSequenceHit.size());
@@ -269,8 +269,8 @@ public List<MatchResult> findMatches(SequenceSketch seqHashes, boolean toSelf)
 
 		//record the search time
 		//TODO not clear why not working. Perhaps everything is too fast?
-		//long endTime = System.nanoTime();
-		//this.sortMergeSearchTime.getAndAdd(endTime-minHashEndTime);
+		long endTime = System.nanoTime();
+		this.sortMergeSearchTime.getAndAdd(endTime-minHashEndTime);
 
 		return matches;
 	}

src/main/java/edu/umd/marbl/mhap/sketch/SequenceSketch.java

@@ -94,7 +94,7 @@ public SequenceSketch(SequenceId id, MinHash mainHashes, OrderKmerHashes ordered
 	}
 
 	public SequenceSketch(Sequence seq, int kmerSize, int numHashes, int orderedKmerSize, 
-			boolean storeHashes, HashSet<Integer> filter, KmerCounts kmerCount)
+			boolean storeHashes, HashSet<Long> filter, KmerCounts kmerCount)
 	{
 		this.id = seq.getId();
 		this.mainHashes = new MinHash(seq, kmerSize, numHashes, filter, kmerCount);

src/main/java/edu/umd/marbl/mhap/sketch/SequenceSketchStreamer.java

@@ -58,7 +58,7 @@ public class SequenceSketchStreamer
 {
 	private final DataInputStream buffInput;
 	private final FastaData fastaData;
-	private final HashSet<Integer> filter;
+	private final HashSet<Long> filter;
 	private final KmerCounts kmerCounter;
 	private final int kmerSize;
 	private final AtomicLong numberProcessed;
@@ -91,7 +91,7 @@ public SequenceSketchStreamer(String file, int offset) throws FileNotFoundExcept
 	}
 
 	public SequenceSketchStreamer(String file, int kmerSize, int numHashes, int subSequenceSize, int orderedKmerSize,
-			HashSet<Integer> filter, KmerCounts kmerCounter, int offset) throws IOException
+			HashSet<Long> filter, KmerCounts kmerCounter, int offset) throws IOException
 	{
 		this.fastaData = new FastaData(file, offset);
 		this.readingFasta = true;

src/main/java/edu/umd/marbl/mhap/utils/Utils.java

@@ -49,7 +49,6 @@
 import com.google.common.hash.Hashing;
 
 import edu.umd.marbl.mhap.general.Sequence;
-import edu.umd.marbl.mhap.sketch.MinHash;
 
 public final class Utils
 {
@@ -460,7 +459,7 @@ public final static int countLetterInRead(String fasta, String letter, Boolean c
 		return count;
 	}
 
-	public final static HashSet<Integer> createKmerFilter(String fileName, double maxPercent, int kmerSize)
+	public final static HashSet<Long> createKmerFilter(String fileName, double maxPercent, int kmerSize)
 			throws IOException
 	{
 		File file = new File(fileName);
@@ -469,7 +468,7 @@ public final static HashSet<Integer> createKmerFilter(String fileName, double ma
 		try (BufferedReader bf = new BufferedReader(new FileReader(file), BUFFER_BYTE_SIZE);)
 		{
 			// generate hashset
-			ArrayList<Integer> filterArray = new ArrayList<Integer>();
+			ArrayList<Long> filterArray = new ArrayList<Long>();
 
 			String line = bf.readLine();
 			while (line != null)
@@ -485,12 +484,12 @@ public final static HashSet<Integer> createKmerFilter(String fileName, double ma
 				// if greater, add to hashset
 				if (percent > maxPercent)
 				{
-					int[] minHash = MinHash.computeKmerMinHashes(str[0], kmerSize, 0, null);
+					long[] minHash = Utils.computeSequenceHashesLong(str[0], kmerSize);
 
 					if (minHash.length > 1)
 						System.err.println("Warning filter file kmer size larger than setting!");
 
-					for (int val : minHash)
+					for (long val : minHash)
 						filterArray.add(val);
 				}
 				else
@@ -499,7 +498,7 @@ public final static HashSet<Integer> createKmerFilter(String fileName, double ma
 				// read the next line
 				line = bf.readLine();
 			}
-			return new HashSet<Integer>(filterArray);
+			return new HashSet<Long>(filterArray);
 		}
 	}