public String translate(Sequence s, GeneticCode gncode) {
String AA = "";
Sequence n = new Sequence();
n.setSequence(s.getSequence().toUpperCase());
int pos = 0;
boolean found = false;
// --Find start codon
// while(!found&&pos<n.getLen()-2) {
// String start=n.getSequence().substring(pos, pos+3);
// if (gncode.getStart_code().get(start)!=null) {
// AA+=""+gncode.getStart_code().get(start);
// pos+=3;
// found=true;
// }
// pos++;
// }
// --return empty string if not found
// if (!found) {
// Config.log("No Start");
// return AA;
// }
pos = start;
// --Continue
n.Remove(0, pos); // remove till start
String codon = n.Remove(0, 3);
// Config.log(n.getSequence());
while (!codon.isEmpty() && n.getLen() > 0) {
// Config.log(codon+" "+gncode.getCode().get(codon));
String aa = gncode.getCode().get(codon);
// Config.log(aa);
if (aa != null) {
AA += aa;
} else {
if (codon.length() == 3) AA += "-";
}
codon = n.Remove(0, 3);
}
// --Remove ending "*" if found
if (AA.endsWith("*")) {
AA = AA.substring(0, AA.length() - 1);
}
return AA;
}
/**
* We read file as fasta but we group sequences by iteration
*
* @param filename
* @return
*/
@Override
public boolean loadSequences(String filename) {
this.name = filename;
try {
seq.clear(); // Clear the sequence vector and open the file
BufferedReader br = new BufferedReader(new FileReader(new File(filename)));
// VARIABLES
boolean sequenceMode = false;
Sequence tmp = new Sequence(); // Temp sequence
StringBuilder tmpsequence = new StringBuilder(); // Temp sequence string
String stri = ""; // Temp read line
// Read the file in a buffer an parse at the same time
// Process :: We read like a fasta file
Config.log("Reading fasta:" + filename + ":");
int count = 0;
String buffer_iteration = ""; // Counter for iterations;
Pattern iter = Pattern.compile("iterations = ([0-9]*)");
while (br.ready()) {
stri = br.readLine();
Matcher m = iter.matcher(stri);
if (m.find()) {
// --Save current Alignment and add to group
if (!buffer_iteration.isEmpty()) {
// --Replace name
if (getNbSequence() > 0) {
for (Sequence s : getSequences()) {
try {
String sname =
(s.getName().startsWith("AZ") ? s.getName().substring(2) : s.getName());
int original_sequence_id = Integer.valueOf(sname);
InfoSequence info = df.getInfoSequence(original_sequence_id);
// System.out.println(info);
s.setName(info.getName());
s.setGi(info.getGi());
s.setAccession(info.getAccession());
s.setAccession_referee(info.getAccession_referee());
s.setAbbreviate(info.getAbbreviate());
s.setSequence_type(info.getSequence_type());
s.setOriginal_id(original_sequence_id);
// System.out.println(s.getAbbreviate());
} catch (Exception e) {
}
}
this.setName("Iterations_" + m.group(1) + "_" + name);
this.setNote("Created on " + Util.returnCurrentDateAndTime());
this.saveToDatabase();
this.setId(0);
this.getSequences().removeAllElements();
}
buffer_iteration = stri;
}
if (sequenceMode && (stri.equals("") || stri.startsWith(">"))) {
tmp.setSequence(tmpsequence.toString());
tmp.loadInfoFromName();
// Add sequence if not empty
if (tmp.getSequence().length() > 0) seq.add(tmp);
tmp = new Sequence();
tmpsequence = new StringBuilder();
sequenceMode = false;
}
if (sequenceMode) {
tmpsequence.append(stri);
count++;
if (count % 10000 == 0) Config.log("*");
}
if (stri.startsWith(">")) {
// We have a fasta definition
tmp.setName(stri.substring(1)); // remove >
sequenceMode = true;
}
} // end while
// Add last read
if (sequenceMode) {
tmp.setSequence(tmpsequence.toString());
tmp.loadInfoFromName();
if (tmp.getSequence().length() > 0) seq.add(tmp);
tmp = new Sequence();
}
br.close();
if (seq.size() == 0) {
if (isDebug()) Config.log("not fasta...");
return false;
}
}
} catch (Exception e) {
e.printStackTrace();
Config.log("Error with " + filename);
return false;
}
if (isDebug()) Config.log("done");
return true;
}
public String aa_to_dna(Sequence original_dna, Sequence protein, GeneticCode gncode) {
LinkedList<String> codons = new LinkedList<String>();
String AA = "";
String dna = "";
Sequence n = new Sequence();
n.setSequence(original_dna.getSequence().toUpperCase());
int pos = 0;
boolean found = false;
// --Find start codon
// while(!found&&pos<n.getLen()-2) {
// String start=n.getSequence().substring(pos, pos+3);
// if (gncode.getStart_code().get(start)!=null) {
// AA+=""+gncode.getStart_code().get(start);
// codons.add(start);
// pos+=3;
// found=true;
// }
// pos++;
// }
// --return empty string if not found
// if (!found) {
// Config.log("No Start");
// return AA;
// }
pos = start;
// --MAKE THE ORIGINAL TRANSLATION
n.Remove(0, pos); // remove till start
String codon = n.Remove(0, 3);
// --Debug
// Config.log(n.getSequence());
while (!codon.isEmpty() && n.getLen() > 0) {
// Config.log(codon+" "+gncode.getCode().get(codon));
String aa = gncode.getCode().get(codon);
// --Debug
// Config.log(aa);
if (aa != null) {
AA += aa;
codons.add(codon);
} else {
if (codon.length() == 3) AA += "-";
}
codon = n.Remove(0, 3);
}
// --MAKE THE REAL cDNA
for (int i = 0; i < protein.getLen(); i++) {
char c = protein.getSequence().charAt(i);
// Config.log(c);
if (c == '-' || c == '.') {
dna += "---";
} else {
// --remove stop codon
String codon_to_add = codons.poll();
if (codon_to_add != null) {
if (gncode.getCode().get(codon_to_add).equals("*")) codon_to_add = "---";
dna += codon_to_add;
}
}
}
return dna;
}
