variantPeptides = new HashSet<>();
String[] alleles = alls.toUpperCase().split(",");
+ StringBuilder vars = new StringBuilder();
+
for (String allele : alleles)
{
allele = allele.trim().toUpperCase();
@@ -168,30 +215,101 @@ public class MappedFeatures
variantCodon[2] = baseCodon[2];
/*
- * poke variant base into canonical codon
+ * poke variant base into canonical codon;
+ * ignore first 'allele' (canonical base)
*/
- int i = cdsPos == codonPos[0] ? 0 : (cdsPos == codonPos[1] ? 1 : 2);
+ final int i = cdsPos == codonPos[0] ? 0
+ : (cdsPos == codonPos[1] ? 1 : 2);
variantCodon[i] = allele.toUpperCase().charAt(0);
+ if (variantCodon[i] == baseCodon[i])
+ {
+ continue;
+ }
String codon = new String(variantCodon);
String peptide = ResidueProperties.codonTranslate(codon);
- if (fromResidue != peptide.charAt(0))
+ boolean synonymous = toResidue == peptide.charAt(0);
+ StringBuilder var = new StringBuilder();
+ if (synonymous)
+ {
+ /*
+ * synonymous variant notation e.g. c.1062C>A(p.=)
+ */
+ var.append("c.").append(String.valueOf(cdsPos))
+ .append(String.valueOf(baseCodon[i])).append(">")
+ .append(String.valueOf(variantCodon[i]))
+ .append("(p.=)");
+ }
+ else
{
- String to3 = ResidueProperties.STOP.equals(peptide) ? "STOP"
+ /*
+ * missense variant notation e.g. p.Arg355Met
+ */
+ String to3 = ResidueProperties.STOP.equals(peptide) ? "Ter"
: StringUtils.toSentenceCase(
ResidueProperties.aa2Triplet.get(peptide));
- String var = "p." + from3 + fromPosition + to3;
- if (!variantPeptides.contains(peptide)) // duplicate consequence
+ var.append("p.").append(from3).append(String.valueOf(toPosition))
+ .append(to3);
+ }
+ if (!variantPeptides.contains(peptide)) // duplicate consequence
+ {
+ variantPeptides.add(peptide);
+ if (vars.length() > 0)
{
- variantPeptides.add(peptide);
- if (vars.length() > 0)
- {
- vars.append(",");
- }
- vars.append(var);
+ vars.append(",");
}
+ vars.append(var);
}
}
return vars.toString();
}
+
+ /**
+ * Answers the name of the linked sequence holding any mapped features
+ *
+ * @return
+ */
+ public String getLinkedSequenceName()
+ {
+ return featureSequence == null ? null : featureSequence.getName();
+ }
+
+ /**
+ * Answers the mapped ranges (as one or more [start, end] positions) which
+ * correspond to the given [begin, end] range of the linked sequence.
+ *
+ *
+ * Example: MappedFeatures with CDS features mapped to peptide
+ * CDS/200-220 gtc aac TGa acGt att AAC tta
+ * mapped to PEP/6-7 WN by mapping [206, 207, 210, 210, 215, 217] to [6, 7]
+ * getMappedPositions(206, 206) should return [6, 6]
+ * getMappedPositions(200, 214) should return [6, 6]
+ * getMappedPositions(210, 215) should return [6, 7]
+ *
+ *
+ * @param begin
+ * @param end
+ * @return
+ */
+ public int[] getMappedPositions(int begin, int end)
+ {
+ MapList map = mapping.getMap();
+ return mapping.to == featureSequence ? map.getOverlapsInFrom(begin, end)
+ : map.getOverlapsInTo(begin, end);
+ }
+
+ /**
+ * Answers true if the linked features are on coding sequence, false if on
+ * peptide
+ *
+ * @return
+ */
+ public boolean isFromCds()
+ {
+ if (mapping.getMap().getFromRatio() == 3)
+ {
+ return mapping.to != featureSequence;
+ }
+ return mapping.to == featureSequence;
+ }
}