/**
* Add annotations
*
* @param category
* @param contrVoc
* @param sequence
*/
void countAaSequence(String category, String contrVoc, String description, String sequence) {
String key = key(category, contrVoc, description);
CountByType cbt = countAaSequenceByType.getOrCreate(key);
cbt.inc(sequence);
}
/** Show annotations counters in a table */
void analyzeSequenceConservation() {
if (verbose)
Timer.showStdErr(
"Sequence conservation analysis." //
+ "\n\tAA sequence length : "
+ 1 //
+ "\n\tMin AA count : "
+ HIGHLY_CONSERVED_AA_COUNT //
+ "\n\tMin AA conservation : "
+ HIGHLY_CONSERVED_AA_PERCENT //
);
ArrayList<String> keys = new ArrayList<String>();
keys.addAll(countAaSequenceByType.keySet());
Collections.sort(keys);
// Show title
StringBuilder title = new StringBuilder();
for (char aa : GprSeq.AMINO_ACIDS) title.append(aa + "\t");
title.append("\t" + title);
if (verbose)
System.out.println(
"Amino acid regions:\n\tTotal\tMax count\tAvg len\tConservation\tCatergory\tControlled Vocabulary\t"
+ title
+ "\tOther AA sequences:");
// Show AA counts for each 'key'
for (String key : keys) {
long seqLen = 0, totalSeqs = 0, maxCount = 0;
CountByType cbt = countAaSequenceByType.get(key);
long total = cbt.sum();
boolean highlyConservedAaSequence = false;
StringBuilder sb = new StringBuilder();
// For each single amino acid "sequence"
for (char aa : GprSeq.AMINO_ACIDS) {
long count = cbt.get("" + aa);
if (count > 0) {
seqLen += 1 * count;
totalSeqs += count;
maxCount = Math.max(maxCount, count);
sb.append(count);
double perc = ((double) count) / total;
// We estimate that if most AA are the same, then changing this AA can cause a high impact
// in protein coding
if ((perc > HIGHLY_CONSERVED_AA_PERCENT) && (total >= HIGHLY_CONSERVED_AA_COUNT))
highlyConservedAaSequence = true;
}
sb.append("\t");
}
// Sequences of more than one AA
for (String aas : cbt.keySet()) {
long count = cbt.get(aas);
double perc = ((double) count) / total;
if (aas.length() > 1) {
seqLen += aas.length() * count;
totalSeqs += count;
maxCount = Math.max(maxCount, count);
sb.append(String.format("\t" + aas + ":" + count));
if ((perc > HIGHLY_CONSERVED_AA_PERCENT) && (total >= HIGHLY_CONSERVED_AA_COUNT))
highlyConservedAaSequence = true;
}
}
long avgLen = seqLen / totalSeqs;
// Show line
if (verbose)
System.out.println( //
"\t"
+ total //
+ "\t"
+ maxCount //
+ "\t"
+ avgLen //
+ "\t"
+ (highlyConservedAaSequence ? "High" : "") //
+ "\t"
+ key //
+ "\t"
+ sb //
);
// Mark highly conserved
if (highlyConservedAaSequence) {
int count = 0;
for (Marker m : markers) {
NextProt nextProt = (NextProt) m;
if (m.getId().equals(key)) {
nextProt.setHighlyConservedAaSequence(true);
count++;
}
}
if (verbose)
Timer.showStdErr(
"NextProt "
+ count
+ " markers type '"
+ key
+ "' marked as highly conserved AA sequence");
}
}
}