From 4099d2eb7db68eb4dd1a27da0fc355fa0108d735 Mon Sep 17 00:00:00 2001
From: Petr Wudi <petr.wudi@gmail.com>
Date: Fri, 17 Nov 2017 23:56:39 +0100
Subject: [PATCH] Modify SimilarDnaFinder interface and change implementation
according to it, closes #5
---
.../AlignedSequencePair.java | 42 ++++++++++
.../fit/kw/vmm/backend/SimilarDnaFinder.java | 18 +++--
.../kw/vmm/backend/SimilarDnaFinderImpl.java | 76 ++++++++++++++-----
3 files changed, 111 insertions(+), 25 deletions(-)
create mode 100644 src/main/java/edu/cvut/fit/kw/vmm/alignment_solution/AlignedSequencePair.java
diff --git a/src/main/java/edu/cvut/fit/kw/vmm/alignment_solution/AlignedSequencePair.java b/src/main/java/edu/cvut/fit/kw/vmm/alignment_solution/AlignedSequencePair.java
new file mode 100644
index 0000000..3e72c47
--- /dev/null
+++ b/src/main/java/edu/cvut/fit/kw/vmm/alignment_solution/AlignedSequencePair.java
@@ -0,0 +1,42 @@
+package edu.cvut.fit.kw.vmm.alignment_solution;
+
+import edu.cvut.fit.kw.vmm.DnaSequence;
+
+/**
+ * Contains pair of sequences and information how to align them to have mathing part of the sequences on the same
+ * position.
+ */
+public class AlignedSequencePair {
+
+ private final AlignmentSolution alignment;
+ private final DnaSequence seq0;
+ private final DnaSequence seq1;
+
+ public AlignedSequencePair(DnaSequence seq0, DnaSequence seq1, AlignmentSolution alignment) {
+ this.seq0 = seq0;
+ this.seq1 = seq1;
+ this.alignment = alignment;
+ }
+
+ /**
+ * @return information how to align them to have mathing part of the sequences on the same position. It specifies
+ * how to transform first string into the second one.
+ */
+ public AlignmentSolution getAlignment() {
+ return alignment;
+ }
+
+ /**
+ * @return first DNA sequence
+ */
+ public DnaSequence getFirstSequence() {
+ return seq0;
+ }
+
+ /**
+ * @return second DNA sequence
+ */
+ public DnaSequence getSecondSequence() {
+ return seq1;
+ }
+}
diff --git a/src/main/java/edu/cvut/fit/kw/vmm/backend/SimilarDnaFinder.java b/src/main/java/edu/cvut/fit/kw/vmm/backend/SimilarDnaFinder.java
index a2c5c94..bba7750 100644
--- a/src/main/java/edu/cvut/fit/kw/vmm/backend/SimilarDnaFinder.java
+++ b/src/main/java/edu/cvut/fit/kw/vmm/backend/SimilarDnaFinder.java
@@ -1,28 +1,32 @@
package edu.cvut.fit.kw.vmm.backend;
+import edu.cvut.fit.kw.vmm.alignment_solution.AlignedSequencePair;
+
import java.util.List;
/**
- * Class containing methods for finding similiar DNA to the specified one
+ * Class containing methods for finding DNA similar to the specified one
*/
public interface SimilarDnaFinder {
/**
* Finds k similar DNA sequences
- * @param sequence sequences similar to this will be found. It is not case-sensitive
+ * @param sequence sequences similar to this will be found
* @param alignment type of alignment (local/global)
* @param k number of similar sequences to find
- * @return list of k dna sequences
+ * @return list of k dna sequence pairs - first sequence is always the one specified in input, second is the one
+ * found in DNA collection
*/
- List<String> findKSimilar(String sequence, AlignmentType alignment, int k);
+ List<AlignedSequencePair> findKSimilar(String sequence, AlignmentType alignment, int k);
/**
* Finds DNA sequences with similarity to the given string higher than specified threshold
- * @param sequence sequences similar to this will be found. It is not case-sensitive
+ * @param sequence sequences similar to this will be found
* @param alignment type of alignment (local/global)
* @param minSimilarity only sequences with value higher or equal this are included in result
- * @return list of most similar DNA sequences
+ * @return list of most similar DNA sequences - first sequence is always the one specified in input, second is the
+ * one found in DNA collection
*/
- List<String> findSimilarWithRange(String sequence, AlignmentType alignment, double minSimilarity);
+ List<AlignedSequencePair> findSimilarWithRange(String sequence, AlignmentType alignment, double minSimilarity);
}
diff --git a/src/main/java/edu/cvut/fit/kw/vmm/backend/SimilarDnaFinderImpl.java b/src/main/java/edu/cvut/fit/kw/vmm/backend/SimilarDnaFinderImpl.java
index 98118b0..5c65534 100644
--- a/src/main/java/edu/cvut/fit/kw/vmm/backend/SimilarDnaFinderImpl.java
+++ b/src/main/java/edu/cvut/fit/kw/vmm/backend/SimilarDnaFinderImpl.java
@@ -1,40 +1,80 @@
package edu.cvut.fit.kw.vmm.backend;
+import edu.cvut.fit.kw.vmm.DnaSequence;
+import edu.cvut.fit.kw.vmm.alignment_solution.AlignedSequencePair;
import edu.cvut.fit.kw.vmm.alignment_solution.AlignmentSolution;
+import edu.cvut.fit.kw.vmm.backend.sequence_containers.SequenceContainer;
import edu.cvut.fit.kw.vmm.backend.sequences_comparement.NeedlemanWunsch;
+import edu.cvut.fit.kw.vmm.backend.sequences_comparement.SequenceComparator;
-import java.util.ArrayList;
import java.util.LinkedList;
import java.util.List;
+/**
+ * Finds similar DNA sequences in specified container
+ */
public class SimilarDnaFinderImpl implements SimilarDnaFinder {
- List<String> sequences;
- List<AlignmentSolution> similarSequences;
+ private static final String INPUT_SEQUENCE_NAME = "input";
+ private final SequenceContainer container;
/**
- * Creates container of sequences. These sequences are stored in specified files.
- * @param inputFiles paths of files containing DNA sequences. If path points to a directory, it recursively finds
- * all files in it and load as DNS sequence files.
+ * Creates new instance of class for finding similar sequences to the specified one in the sequence container.
+ * @param container object containing all the sequences which will be compared to the specified one
*/
- public SimilarDnaFinderImpl(List<String> inputFiles) {
- sequences = new ArrayList<>();
- similarSequences = new LinkedList<>();
+ public SimilarDnaFinderImpl(SequenceContainer container) {
+ this.container = container;
}
@Override
- public List<String> findKSimilar(String sequence, AlignmentType alignment, int k) {
- List<String> best;
- for(String refSeq : sequences) {
- NeedlemanWunsch solver = new NeedlemanWunsch(refSeq, sequence);
- AlignmentSolution solution = solver.solve();
- // TODO: saving the best results
+ public List<AlignedSequencePair> findKSimilar(String inputSeqString, AlignmentType alignmentType, int k) {
+ DnaSequence inputSequence = new DnaSequence(INPUT_SEQUENCE_NAME, inputSeqString);
+ List<AlignedSequencePair> similarSequences = new LinkedList<>(); // First has lowest similarity, last highest
+ for(DnaSequence sequence : container) {
+ String seqString = sequence.getSequence();
+ SequenceComparator comparator = null;
+ if(alignmentType == AlignmentType.GLOBAL) {
+ comparator = new NeedlemanWunsch(inputSeqString, seqString);
+ AlignmentSolution alignment = comparator.solve();
+ tryFitInList(inputSequence, sequence, alignment, similarSequences, k);
+ }
+ }
+ return similarSequences;
+ }
+
+ private void tryFitInList(DnaSequence seq0,
+ DnaSequence seq1,
+ AlignmentSolution alignment,
+ List<AlignedSequencePair> similarSequences,
+ int maxSize) {
+ int firstHigherScoreValue = 0; // First sequence in the list which has score higher than this one
+ while(firstHigherScoreValue < similarSequences.size() &&
+ similarSequences.get(firstHigherScoreValue).getAlignment().getSimilarity() < alignment.getSimilarity()) {
+ firstHigherScoreValue++;
+ }
+ if(firstHigherScoreValue > 0 || similarSequences.size() < maxSize) {
+ similarSequences.add(firstHigherScoreValue, new AlignedSequencePair(seq0, seq1, alignment));
+ if(similarSequences.size() > maxSize) {
+ similarSequences.remove(0);
+ }
}
- return null; // TODO
}
@Override
- public List<String> findSimilarWithRange(String sequence, AlignmentType alignment, double minSimilarity) {
- return null;
+ public List<AlignedSequencePair> findSimilarWithRange(String inputSeqString, AlignmentType alignmentType, double minSimilarity) {
+ DnaSequence inputSequence = new DnaSequence(INPUT_SEQUENCE_NAME, inputSeqString);
+ List<AlignedSequencePair> similarSequences = new LinkedList<>(); // First has lowest similarity, last highest
+ for(DnaSequence sequence : container) {
+ String seqString = sequence.getSequence();
+ SequenceComparator comparator = null;
+ if(alignmentType == AlignmentType.GLOBAL) {
+ comparator = new NeedlemanWunsch(inputSeqString, seqString);
+ AlignmentSolution alignment = comparator.solve();
+ if(alignment.getSimilarity() >= minSimilarity) {
+ similarSequences.add(new AlignedSequencePair(inputSequence, sequence, alignment));
+ }
+ }
+ }
+ return similarSequences;
}
}
--
GitLab