public static ChromosomeArray regenNBlocks(
ChromosomeArray cha, int blocksize, int trigger, int endsize) {
ChromosomeArray chb =
new ChromosomeArray(cha.chromosome, cha.strand, cha.minIndex, cha.maxIndex);
chb.maxIndex = -1;
int aloc = 0;
int bloc = 0;
int ns = 0;
// Process start
while (cha.get(aloc) == 'N') {
chb.set(bloc, 'N');
ns++;
aloc++;
bloc++;
}
while (ns < endsize) {
chb.set(bloc, 'N');
ns++;
bloc++;
}
ns = 0;
// Process middle
while (aloc <= cha.maxIndex) {
byte b = cha.get(aloc);
if (b == 'N') {
ns++;
} else {
if (ns >= trigger) {
while (ns < blocksize) {
chb.set(bloc, 'N');
bloc++;
ns++;
}
}
ns = 0;
}
chb.set(bloc, b);
aloc++;
bloc++;
}
// Process end
ns = 0;
for (int i = chb.maxIndex; i >= 0; i--) {
if (chb.get(i) != 'N') {
break;
}
}
while (ns < endsize) {
chb.set(chb.maxIndex + 1, 'N');
ns++;
}
return chb;
}
/**
* @param replaceFirst
* @param chromFname
* @param chrom
*/
public static void process(String inVarsName, String outChromName, int chrom) {
ArrayList<Varlet> vars = Varlet.fromTextFile(inVarsName);
ChromosomeArray cha = Data.getChromosome(chrom);
ChromosomeArray chb = new ChromosomeArray(chrom, Gene.PLUS);
// Next location to read in a
int aloc = 0;
// Next location to set in b
int bloc = 0;
for (int i = 0; i < vars.size(); i++) {
Varlet v = vars.get(i);
assert (v.beginLoc >= aloc)
: i + "\n" + vars.get(i - 1) + "\n" + v + "\n"; // Overlapping variations
while (v.beginLoc < aloc) { // skip it, for now.
System.err.print("e");
i++;
if (i >= vars.size()) {
break;
}
v = vars.get(i);
}
if (STAY_NEAR_REF && Tools.absdif(aloc, bloc) >= REF_LIMIT) {
int dif = v.lengthDif();
if (aloc < bloc) { // skip insertions
while (dif > 0) {
// System.err.print("i");
i++;
if (i >= vars.size()) {
break;
}
v = vars.get(i);
dif = v.lengthDif();
}
} else { // skip deletions
while (dif < 0) {
// System.err.print("d");
i++;
if (i >= vars.size()) {
break;
}
v = vars.get(i);
dif = v.lengthDif();
}
}
}
// Advance to variation's beginning
while (aloc < v.beginLoc) {
byte b = cha.get(aloc);
chb.set(bloc, b);
aloc++;
bloc++;
}
// Apply variation
if (v.varType == Variation.SNP) {
String call = v.call;
String ref = v.ref;
if (ref != null && ref.equals("=")) {
ref = null;
}
for (int j = 0; j < call.length(); j++) {
char c = call.charAt(j);
if (ref != null) {
assert (ref.charAt(j) == cha.get(aloc)) : "\n" + i + ", " + v;
}
chb.set(bloc, c);
aloc++;
bloc++;
}
} else if (v.varType == Variation.DELINS) {
String call = v.call;
for (int j = 0; j < call.length(); j++) {
char c = call.charAt(j);
chb.set(bloc, c);
bloc++;
}
aloc += v.lengthRef();
} else if (v.varType == Variation.NOCALL) {
// Do nothing. But, it should have been removed already.
if (!foundNocall) {
System.err.println("*** Warning - found a nocall in input variations ***");
foundNocall = true;
}
} else if (v.varType == Variation.NOREF) {
String call = v.call;
for (int j = 0; j < call.length(); j++) {
char c = call.charAt(j);
assert (cha.get(aloc) == 'N') : cha.get(aloc);
chb.set(bloc, c);
aloc++;
bloc++;
}
} else if (v.varType == Variation.INS) {
String call = v.call;
for (int j = 0; j < call.length(); j++) {
char c = call.charAt(j);
chb.set(bloc, c);
bloc++;
}
} else if (v.varType == Variation.DEL) {
int len = v.lengthRef();
assert (len > 0);
aloc += len;
}
}
// Finish writing array
while (aloc < cha.array.length || aloc <= cha.maxIndex) {
byte c = cha.get(aloc);
chb.set(bloc, c);
aloc++;
bloc++;
}
System.out.println("Length Shift for chr" + chrom + ": \t" + (bloc - aloc));
Data.unload(chrom, true);
cha = null;
if (REGEN_N_BLOCKS) {
chb = regenNBlocks(chb, N_BLOCK_SIZE, N_BLOCK_TRIGGER, N_BLOCK_END_SIZE);
}
chb.resize(chb.maxIndex + 1);
// Can't do this because it is read later
// if(THREADS==1){ReadWrite.writeObjectInThread(cac, outChromName);}
// else{ReadWrite.write(cac, outChromName);}
ReadWrite.write(chb, outChromName, false);
}