private int[][] getPwmByName(String tfbsName, ArrayList<RegulatoryElementPWM> pwms) {
for (Iterator<RegulatoryElementPWM> iterator = pwms.iterator(); iterator.hasNext(); ) {
RegulatoryElementPWM tfbsPWM = (RegulatoryElementPWM) iterator.next();
if (tfbsPWM.getName().equals(tfbsName)) {
return tfbsPWM.getPwm();
}
}
return null;
}
// TODO: faze it out - now I use Hash as above. Make sure that the score
public ArrayList<RegulatoryElementPWM> getRegulatoryElementsPWMs() throws DataFormatException {
String pwmLineRegEx1 = "[ACGTacgt][ \t]*\\|([ \t]*\\d+)+[ \t]*";
String pwmLineRegEx2 = "[ACGTacgt][ \t]*\\[([ \t]*\\d+)+[ \t]*\\][ \t]*";
ArrayList<RegulatoryElementPWM> pwms = new ArrayList<RegulatoryElementPWM>();
// *** Get file names of all matrixes
final String[] matrixFileNames = new File(matrixFilesDir).list();
if (matrixFileNames == null) {
throw new DataFormatException(
"Patser Regulatory Element Service has no PSSMs associated with it.");
}
for (int i = 0; i < matrixFileNames.length; i++) {
// Only read files with .matrix extension
if (!matrixFileNames[i].endsWith(SystemVariables.getInstance().getString("pwm.extension"))) {
continue;
}
RegulatoryElementPWM currPwmObj = new RegulatoryElementPWM();
String tfbsName = matrixFileNames[i].substring(0, matrixFileNames[i].length() - 7);
currPwmObj.setName(tfbsName);
BufferedReader bufferedReader = null;
try {
// for Mac
bufferedReader =
new BufferedReader(new FileReader(matrixFilesDir + "/" + matrixFileNames[i]));
// bufferedReader = new BufferedReader(new FileReader(matrixFilesDir + matrixFileNames[i]));
String line = null;
int[][] pwmArr = null;
int j = -1;
while (null != (line = bufferedReader.readLine())) {
if (line.matches(pwmLineRegEx1) || line.matches(pwmLineRegEx2)) {
j++;
if (j > 3)
throw new DataFormatException(
"Error parsing matrix file <"
+ tfbsName
+ ".matrix>. Unexpected line in the file. File should contain only one matrix.");
line = line.replaceAll("[ACGTacgt\\|\\[\\]]", ""); // remove everything but the numbers
StringTokenizer strTok = new StringTokenizer(line);
int lengthOfPwm = strTok.countTokens();
if (pwmArr == null) {
pwmArr = new int[lengthOfPwm][4];
}
if (lengthOfPwm != pwmArr.length)
throw new DataFormatException(
"Error parsing matrix file <" + tfbsName + ".matrix>. Matrix is unbalanced.");
int k = -1;
while (strTok.hasMoreElements()) {
k++;
String token = strTok.nextToken();
pwmArr[k][j] = Integer.parseInt(token);
}
}
}
currPwmObj.setPwm(pwmArr);
} catch (FileNotFoundException ex) {
ex.printStackTrace();
} catch (IOException ex) {
ex.printStackTrace();
} finally {
try {
if (bufferedReader != null) bufferedReader.close();
} catch (IOException ex) {
ex.printStackTrace();
}
}
pwms.add(currPwmObj);
} // matrix for
return pwms;
}
