|
| 1 | +package org.biojava.nbio.phylo; |
| 2 | + |
| 3 | +import java.io.ByteArrayOutputStream; |
| 4 | +import java.io.InputStream; |
| 5 | +import java.io.OutputStream; |
| 6 | +import java.util.LinkedHashMap; |
| 7 | + |
| 8 | +import org.biojava.nbio.core.sequence.MultipleSequenceAlignment; |
| 9 | +import org.biojava.nbio.core.sequence.ProteinSequence; |
| 10 | +import org.biojava.nbio.core.sequence.compound.AminoAcidCompound; |
| 11 | +import org.biojava.nbio.core.sequence.compound.AminoAcidCompoundSet; |
| 12 | +import org.biojava.nbio.core.sequence.io.FastaReader; |
| 13 | +import org.biojava.nbio.core.sequence.io.FastaWriter; |
| 14 | +import org.biojava.nbio.core.sequence.io.GenericFastaHeaderParser; |
| 15 | +import org.biojava.nbio.core.sequence.io.ProteinSequenceCreator; |
| 16 | +import org.biojava.nbio.core.sequence.io.template.FastaHeaderFormatInterface; |
| 17 | +import org.forester.msa.Msa; |
| 18 | +import org.junit.Test; |
| 19 | + |
| 20 | +import static org.junit.Assert.*; |
| 21 | + |
| 22 | +/** |
| 23 | + * Test the BioJava-forester wrapper methods. |
| 24 | + * |
| 25 | + * @author Aleix Lafita |
| 26 | + * |
| 27 | + */ |
| 28 | +public class TestForesterWrapper { |
| 29 | + |
| 30 | + @Test |
| 31 | + public void testMSAconversion() throws Exception { |
| 32 | + |
| 33 | + // Load the msa FASTA file into a BioJava MSA object |
| 34 | + InputStream inStream = TestForesterWrapper.class |
| 35 | + .getResourceAsStream("/1u6d_symm.fasta"); |
| 36 | + |
| 37 | + FastaReader<ProteinSequence, AminoAcidCompound> fastaReader = |
| 38 | + new FastaReader<ProteinSequence, AminoAcidCompound>( |
| 39 | + inStream, |
| 40 | + new GenericFastaHeaderParser<ProteinSequence, AminoAcidCompound>(), |
| 41 | + new ProteinSequenceCreator(AminoAcidCompoundSet |
| 42 | + .getAminoAcidCompoundSet())); |
| 43 | + |
| 44 | + LinkedHashMap<String, ProteinSequence> proteinSequences = fastaReader |
| 45 | + .process(); |
| 46 | + |
| 47 | + inStream.close(); |
| 48 | + |
| 49 | + MultipleSequenceAlignment<ProteinSequence, AminoAcidCompound> msa = |
| 50 | + new MultipleSequenceAlignment<ProteinSequence, AminoAcidCompound>(); |
| 51 | + |
| 52 | + String expected = ""; |
| 53 | + for (ProteinSequence proteinSequence : proteinSequences.values()) { |
| 54 | + msa.addAlignedSequence(proteinSequence); |
| 55 | + expected += ">" + proteinSequence.getOriginalHeader() + "\n" |
| 56 | + + proteinSequence.toString() + "\n"; |
| 57 | + } |
| 58 | + |
| 59 | + // Convert the biojava MSA to a FASTA String |
| 60 | + OutputStream os = new ByteArrayOutputStream(); |
| 61 | + FastaWriter<ProteinSequence, AminoAcidCompound> fastaW = |
| 62 | + new FastaWriter<ProteinSequence, AminoAcidCompound>(os, |
| 63 | + msa.getAlignedSequences(), |
| 64 | + new FastaHeaderFormatInterface<ProteinSequence, AminoAcidCompound>() { |
| 65 | + @Override |
| 66 | + public String getHeader(ProteinSequence sequence) { |
| 67 | + return sequence.getAccession().toString(); |
| 68 | + }; |
| 69 | + }); |
| 70 | + fastaW.process(); |
| 71 | + String biojava = os.toString(); |
| 72 | + |
| 73 | + // Convert the biojava MSA to a forester Msa |
| 74 | + Msa fMsa = ForesterWrapper.convert(msa); |
| 75 | + |
| 76 | + StringBuilder sb = new StringBuilder(); |
| 77 | + for (int i = 0; i < fMsa.getNumberOfSequences(); i++) { |
| 78 | + sb.append(">" + fMsa.getIdentifier(i) + "\n"); |
| 79 | + sb.append(fMsa.getSequenceAsString(i) + "\n"); |
| 80 | + } |
| 81 | + String forester = sb.toString(); |
| 82 | + |
| 83 | + // Assert that all FASTA files are equal |
| 84 | + assertEquals(expected, biojava); |
| 85 | + assertEquals(expected, forester); |
| 86 | + |
| 87 | + } |
| 88 | +} |
0 commit comments