/**
* 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);
}
