@Test
public void testSmallFasta2() {
try {
InputStream inStream = this.getClass().getResourceAsStream("/test.fasta");
FastaReader<ProteinSequence, AminoAcidCompound> fastaReader =
new FastaReader<ProteinSequence, AminoAcidCompound>(
inStream,
new GenericFastaHeaderParser<ProteinSequence, AminoAcidCompound>(),
new ProteinSequenceCreator(AminoAcidCompoundSet.getAminoAcidCompoundSet()));
int nrSeq = 0;
LinkedHashMap<String, ProteinSequence> b = fastaReader.process();
assertNotNull(b);
// #282 make sure that process() still works
assertTrue(b.keySet().size() == 10);
} catch (Exception ex) {
ex.printStackTrace();
java.util.logging.Logger.getLogger(this.getClass().getName()).log(Level.SEVERE, null, ex);
fail(ex.getMessage());
}
}
@Test
public void testSimpleSubstitutionMatrixCompoundSetOfCShortShort() {
SubstitutionMatrix<AminoAcidCompound> matrix =
new SimpleSubstitutionMatrix<AminoAcidCompound>(
AminoAcidCompoundSet.getAminoAcidCompoundSet(), (short) 5, (short) 1);
assertEquals(matrix.getName(), "IDENTITY_5_1");
}
@Test
public void testSmallFasta() {
try {
InputStream inStream = this.getClass().getResourceAsStream("/test.fasta");
FastaReader<ProteinSequence, AminoAcidCompound> fastaReader =
new FastaReader<ProteinSequence, AminoAcidCompound>(
inStream,
new GenericFastaHeaderParser<ProteinSequence, AminoAcidCompound>(),
new ProteinSequenceCreator(AminoAcidCompoundSet.getAminoAcidCompoundSet()));
LinkedHashMap<String, ProteinSequence> b;
int nrSeq = 0;
while ((b = fastaReader.process(10)) != null) {
for (String key : b.keySet()) {
nrSeq++;
// #282 would result in an endless loop
// this makes sure it has been fixed.
assertTrue(
"Looks like there is a problem with termination of processing of the FASTA file!",
nrSeq < 15);
}
}
} catch (Exception ex) {
ex.printStackTrace();
java.util.logging.Logger.getLogger(this.getClass().getName()).log(Level.SEVERE, null, ex);
fail(ex.getMessage());
}
}
@Test()
public void testSimpleSubstitutionMatrix() {
SubstitutionMatrix<AminoAcidCompound> matrix = SimpleSubstitutionMatrix.getBlosum62();
assertEquals(matrix.getCompoundSet(), AminoAcidCompoundSet.getAminoAcidCompoundSet());
assertEquals(matrix.getName(), "blosum62");
assertEquals(matrix.getMaxValue(), 11);
assertEquals(matrix.getMinValue(), -4);
}
/**
* Reads the file {@code fastaFile}, expecting exactly two sequences which give a pairwise
* alignment. Uses this and two structures to create an AFPChain corresponding to the alignment.
* Uses a {@link CasePreservingProteinSequenceCreator} and assumes that a residue is aligned if
* and only if it is given by an uppercase letter.
*
* @see #fastaToAfpChain(ProteinSequence, ProteinSequence, Structure, Structure)
* @throws IOException
* @throws StructureException
*/
public static AFPChain fastaFileToAfpChain(
File fastaFile, Structure structure1, Structure structure2)
throws IOException, StructureException {
InputStream inStream = new FileInputStream(fastaFile);
SequenceCreatorInterface<AminoAcidCompound> creator =
new CasePreservingProteinSequenceCreator(AminoAcidCompoundSet.getAminoAcidCompoundSet());
SequenceHeaderParserInterface<ProteinSequence, AminoAcidCompound> headerParser =
new GenericFastaHeaderParser<ProteinSequence, AminoAcidCompound>();
FastaReader<ProteinSequence, AminoAcidCompound> fastaReader =
new FastaReader<ProteinSequence, AminoAcidCompound>(inStream, headerParser, creator);
LinkedHashMap<String, ProteinSequence> sequences = fastaReader.process();
inStream.close();
return fastaToAfpChain(sequences, structure1, structure2);
}
/**
* Takes a structure and sequence corresponding to an alignment between a structure or sequence
* and itself (or even a structure with a sequence), where the result has a circular permutation
* site {@link cpSite} residues to the right.
*
* @param fastaFile A FASTA file containing exactly 2 sequences, the first unpermuted and the
* second permuted
* @param cpSite The number of residues from the beginning of the sequence at which the circular
* permutation site occurs; can be positive or negative; values greater than the length of the
* sequence are acceptable
* @throws IOException
* @throws StructureException
*/
public static AFPChain cpFastaToAfpChain(File fastaFile, Structure structure, int cpSite)
throws IOException, StructureException {
InputStream inStream = new FileInputStream(fastaFile);
SequenceCreatorInterface<AminoAcidCompound> creator =
new CasePreservingProteinSequenceCreator(AminoAcidCompoundSet.getAminoAcidCompoundSet());
SequenceHeaderParserInterface<ProteinSequence, AminoAcidCompound> headerParser =
new GenericFastaHeaderParser<ProteinSequence, AminoAcidCompound>();
FastaReader<ProteinSequence, AminoAcidCompound> fastaReader =
new FastaReader<ProteinSequence, AminoAcidCompound>(inStream, headerParser, creator);
LinkedHashMap<String, ProteinSequence> sequences = fastaReader.process();
inStream.close();
Iterator<ProteinSequence> iter = sequences.values().iterator();
ProteinSequence first = iter.next();
ProteinSequence second = iter.next();
return cpFastaToAfpChain(first, second, structure, cpSite);
}
/** Test of process method, of class FastaReader. */
@Test
public void testProcess() throws Exception {
logger.info("process");
InputStream inStream = this.getClass().getResourceAsStream("/PF00104_small.fasta");
assertNotNull(inStream);
FastaReader<ProteinSequence, AminoAcidCompound> fastaReader =
new FastaReader<ProteinSequence, AminoAcidCompound>(
inStream,
new GenericFastaHeaderParser<ProteinSequence, AminoAcidCompound>(),
new ProteinSequenceCreator(AminoAcidCompoundSet.getAminoAcidCompoundSet()));
LinkedHashMap<String, ProteinSequence> proteinSequences = fastaReader.process();
inStream.close();
// Should have 282 sequences
// logger.debug("Expecting 283 got " + proteinSequences.size());
assertEquals(proteinSequences.size(), 283);
int seqNum = 0;
for (String id : proteinSequences.keySet()) {
ProteinSequence proteinSequence = proteinSequences.get(id);
switch (seqNum) {
case 0:
assertEquals(proteinSequence.getAccession().getID(), "A2D504_ATEGE/1-46");
assertEquals(
proteinSequence.getSequenceAsString(),
"-----------------FK-N----LP-LED----------------Q----ITL--IQY-----------SWM----------------------CL-SSFA------LSWRSYK---HTNSQFLYFAPDLVF-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------");
break;
case 281:
// logger.debug("Get Accession: {}", proteinSequence.getAccession());
// logger.debug("Get Protein Sequence: {}", proteinSequence.getSequenceAsString());
assertEquals(proteinSequence.getAccession().getID(), "Q9PU76_CRONI/141-323");
assertEquals(
proteinSequence.getSequenceAsString(),
"VETVTELTEFAKSI-PGFS-N----LD-LND----------------Q----VTL--LKY-----------GVY----------------------EA-IFAM------LASVMNK---DGMPVAYGNGFITRE------------------------------------------------------------------------------------------------------------------------------------------------------------FLKSLRKPFCDIMEPKFDFA-MKF-NSL-E-LDDSDI--------------------SLFVA-AIIC-CGDRPG-------------------------------------------LVNV--GHIEKMQESIVHVLKL-H-----LQN---------NH---PD----------------------------DI------F--------LFP-KLLQKMAD-LRQLV-----------------TEH-AQLV--QIIKK---TESDAHLHPLL-------QEI---");
break;
case 282:
assertEquals(proteinSequence.getAccession().getID(), "Q98SJ1_CHICK/15-61");
assertEquals(
proteinSequence.getSequenceAsString(),
"---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------Q-----------------NW------Q--------RFY-QLTKLLDS-MHDVV-----------------ENL-LSFC--FQTFLDKSM--SIEFPEML-------AEI---");
break;
}
seqNum++;
}
assertEquals(seqNum, 283);
}
/** Test of process method, of class GenbankReader. */
@Test
public void testProcess() throws Throwable {
/*
* Method 1: With the GenbankProxySequenceReader
*/
// Try with the GenbankProxySequenceReader
GenbankProxySequenceReader<AminoAcidCompound> genbankProteinReader =
new GenbankProxySequenceReader<AminoAcidCompound>(
System.getProperty("java.io.tmpdir"),
"NP_000257",
AminoAcidCompoundSet.getAminoAcidCompoundSet());
ProteinSequence proteinSequence = new ProteinSequence(genbankProteinReader);
genbankProteinReader
.getHeaderParser()
.parseHeader(genbankProteinReader.getHeader(), proteinSequence);
logger.info(
"Sequence({},{}) = {}...",
proteinSequence.getAccession(),
proteinSequence.getLength(),
proteinSequence.getSequenceAsString().substring(0, 10));
GenbankProxySequenceReader<NucleotideCompound> genbankDNAReader =
new GenbankProxySequenceReader<NucleotideCompound>(
System.getProperty("java.io.tmpdir"), "NM_001126", DNACompoundSet.getDNACompoundSet());
DNASequence dnaSequence = new DNASequence(genbankDNAReader);
genbankDNAReader.getHeaderParser().parseHeader(genbankDNAReader.getHeader(), dnaSequence);
logger.info(
"Sequence({},{}) = {}...",
dnaSequence.getAccession(),
dnaSequence.getLength(),
dnaSequence.getSequenceAsString().substring(0, 10));
/*
* Method 2: With the GenbankReaderHelper
*/
// Try with the GenbankReaderHelper
ClasspathResource dnaResource = new ClasspathResource("NM_000266.gb", true);
// File dnaFile = new File("src/test/resources/NM_000266.gb");
// File protFile = new File("src/test/resources/BondFeature.gb");
ClasspathResource protResource = new ClasspathResource("BondFeature.gb");
LinkedHashMap<String, DNASequence> dnaSequences =
GenbankReaderHelper.readGenbankDNASequence(dnaResource.getInputStream());
for (DNASequence sequence : dnaSequences.values()) {
logger.info("DNA Sequence: {}", sequence.getSequenceAsString());
}
LinkedHashMap<String, ProteinSequence> protSequences =
GenbankReaderHelper.readGenbankProteinSequence(protResource.getInputStream());
for (ProteinSequence sequence : protSequences.values()) {
logger.info("Protein Sequence: {}", sequence.getSequenceAsString());
}
/*
* Method 3: With the GenbankReader Object
*/
// Try reading with the GanbankReader
GenbankReader<DNASequence, NucleotideCompound> dnaReader =
new GenbankReader<DNASequence, NucleotideCompound>(
dnaResource.getInputStream(),
new GenericGenbankHeaderParser<DNASequence, NucleotideCompound>(),
new DNASequenceCreator(DNACompoundSet.getDNACompoundSet()));
dnaSequences = dnaReader.process();
logger.info("DNA Sequence: {}", dnaSequences);
GenbankReader<ProteinSequence, AminoAcidCompound> protReader =
new GenbankReader<ProteinSequence, AminoAcidCompound>(
protResource.getInputStream(),
new GenericGenbankHeaderParser<ProteinSequence, AminoAcidCompound>(),
new ProteinSequenceCreator(AminoAcidCompoundSet.getAminoAcidCompoundSet()));
protSequences = protReader.process();
logger.info("Protein Sequence: {}", protSequences);
}
@Test
public void processIntTest() throws Exception {
logger.info("process(int)");
InputStream inStream = this.getClass().getResourceAsStream("/PF00104_small.fasta");
assertNotNull(inStream);
FastaReader<ProteinSequence, AminoAcidCompound> fastaReader =
new FastaReader<ProteinSequence, AminoAcidCompound>(
inStream,
new GenericFastaHeaderParser<ProteinSequence, AminoAcidCompound>(),
new ProteinSequenceCreator(AminoAcidCompoundSet.getAminoAcidCompoundSet()));
LinkedHashMap<String, ProteinSequence> proteinSequences = fastaReader.process(200);
// Should have 200 sequences
// logger.debug("Expecting 200 got " + proteinSequences.size());
assertEquals(proteinSequences.size(), 200);
int seqNum = 0;
for (String id : proteinSequences.keySet()) {
ProteinSequence proteinSequence = proteinSequences.get(id);
switch (seqNum) {
case 0:
assertEquals(proteinSequence.getAccession().getID(), "A2D504_ATEGE/1-46");
assertEquals(
proteinSequence.getSequenceAsString(),
"-----------------FK-N----LP-LED----------------Q----ITL--IQY-----------SWM----------------------CL-SSFA------LSWRSYK---HTNSQFLYFAPDLVF-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------");
break;
case 199:
assertEquals(proteinSequence.getAccession().getID(), "Q5F0P7_HUMAN/248-428");
assertEquals(
proteinSequence.getSequenceAsString(),
"DRELVVIIGWAKHI-PGFS-S----LS-LGD----------------Q----MSL--LQS-----------AWM----------------------EI-LILG------IVYRSLP---YDDKLVYAEDYIMD-------------------------------------------------------------------------------------------------------------------------------------------------------------EEHSRLAGLLELYRAILQLV-RRY-KKL-K-VEKEEF--------------------VTLKA-LALA-NSDSMY-------------------------------------------IEDL--EAVQKLQDLLHEALQD-Y-----ELS---------QR---HE----------------------------EP------W--------RTG-KLLLTLPL-LRQTA-----------------AKA-VQHF--YSVKLQGKV--PMH--KLF-------LEM---");
break;
}
seqNum++;
}
assertEquals(seqNum, 200);
// Should have 83 sequences
proteinSequences = fastaReader.process(200);
assertEquals(proteinSequences.size(), 83);
seqNum = 0;
for (String id : proteinSequences.keySet()) {
ProteinSequence proteinSequence = proteinSequences.get(id);
switch (seqNum) {
case 0:
assertEquals(proteinSequence.getAccession().getID(), "RARA_CANFA/233-413");
assertEquals(
proteinSequence.getSequenceAsString(),
"TKCIIKTVEFAKQL-PGFT-T----LT-IAD----------------Q----ITL--LKA-----------ACL----------------------DI-LILR------ICTRYTP---EQDTMTFSEGLTLN-------------------------------------------------------------------------------------------------------------------------------------------------------------RTQMHKAGFGPLTDLVFAFA-NQL-LPL-E-MDDAET--------------------GLLSA-ICLI-CGDRQD-------------------------------------------LEQP--DRVDMLQEPLLEALKV-Y-----VRK---------RR---PS----------------------------RP------H--------MFP-KMLMKITD-LRSIS-----------------AKG-AERV--ITLKMEIPG--SMP--PLI-------QEM---");
break;
case 81:
// logger.debug(proteinSequence.getAccession());
// logger.debug(proteinSequence.getSequenceAsString());
assertEquals(proteinSequence.getAccession().getID(), "Q9PU76_CRONI/141-323");
assertEquals(
proteinSequence.getSequenceAsString(),
"VETVTELTEFAKSI-PGFS-N----LD-LND----------------Q----VTL--LKY-----------GVY----------------------EA-IFAM------LASVMNK---DGMPVAYGNGFITRE------------------------------------------------------------------------------------------------------------------------------------------------------------FLKSLRKPFCDIMEPKFDFA-MKF-NSL-E-LDDSDI--------------------SLFVA-AIIC-CGDRPG-------------------------------------------LVNV--GHIEKMQESIVHVLKL-H-----LQN---------NH---PD----------------------------DI------F--------LFP-KLLQKMAD-LRQLV-----------------TEH-AQLV--QIIKK---TESDAHLHPLL-------QEI---");
break;
case 82:
assertEquals(proteinSequence.getAccession().getID(), "Q98SJ1_CHICK/15-61");
assertEquals(
proteinSequence.getSequenceAsString(),
"---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------Q-----------------NW------Q--------RFY-QLTKLLDS-MHDVV-----------------ENL-LSFC--FQTFLDKSM--SIEFPEML-------AEI---");
break;
}
seqNum++;
}
assertEquals(seqNum, 83);
fastaReader.close();
inStream.close();
}
@Test(expected = FileNotFoundException.class)
public void testSimpleSubstitutionMatrixNotFound() throws FileNotFoundException {
new SimpleSubstitutionMatrix<AminoAcidCompound>(
AminoAcidCompoundSet.getAminoAcidCompoundSet(), new File("blosum63.txt"));
}