Variant at CDS end, test cases added for HGVS notation

src/main/java/org/snpeff/codons/CodonTable.java

@@ -13,6 +13,9 @@
  */
 public class CodonTable {
 
+	public static final String TERMINATION_CODON = "Ter";
+	public static final String TERMINATION_CODON_1 = "*";
+
 	private static HashMap<String, String> aa3letter;
 
 	/**
@@ -108,7 +111,7 @@ public String aa(String codons, boolean fullProteinSequence) {
 	}
 
 	public String aaThreeLetterCode(char aa) {
-		if (aa == '*') return "Ter"; // Termination codon. Used to be "*" (see reference http://www.hgvs.org/mutnomen/standards.html#aalist)
+		if (aa == '*') return TERMINATION_CODON; // Termination codon. Used to be "*" (see reference http://www.hgvs.org/mutnomen/standards.html#aalist)
 		String aa3 = aa3letter.get(Character.toString(aa).toUpperCase());
 		if (aa3 == null) return "???";
 		return aa3;

src/main/java/org/snpeff/interval/Variant.java

@@ -411,7 +411,7 @@ public int lengthChange() {
 	 * Return the change (always compared to 'referenceStrand')
 	 */
 	public String netChange(boolean reverseStrand) {
-		if (isDel()) return reverseStrand ? GprSeq.reverseWc(ref) : ref; // Deleteion have empty 'alt'
+		if (isDel()) return reverseStrand ? GprSeq.reverseWc(ref) : ref; // Deletions have empty 'alt'
 		return reverseStrand ? GprSeq.reverseWc(alt) : alt; // Need reverse-WC?
 	}
 

src/main/java/org/snpeff/interval/codonChange/CodonChange.java

@@ -173,7 +173,7 @@ public void codonChange() {
 
 				if (cdsBaseInExon < 0) cdsBaseInExon = 0;
 
-				// Get codon number and index within codon (where seqChage is pointing)
+				// Get codon number and index within codon
 				if (codonStartNum < 0) {
 					codonStartNum = (firstCdsBaseInExon + cdsBaseInExon) / CODON_SIZE;
 					codonStartIndex = (firstCdsBaseInExon + cdsBaseInExon) % CODON_SIZE;
@@ -242,6 +242,13 @@ protected String codonsRef(int numCodons) {
 		return codon;
 	}
 
+	/**
+	 * Calculate variant effect
+	 * @param marker: Genomic marker affected by this variant (e.g. exon, transcript, etc)
+	 * @param effectType: Effect type
+	 * @param allowReplace: Can another variant effect replace this one?
+	 * @return A new VariantEffect object
+	 */
 	protected VariantEffect effect(Marker marker, EffectType effectType, boolean allowReplace) {
 		return effect(marker, effectType, effectType.effectImpact(), codonsRef, codonsAlt, codonStartNum, codonStartIndex, allowReplace);
 	}

src/main/java/org/snpeff/interval/codonChange/CodonChangeIns.java

@@ -17,6 +17,19 @@ public CodonChangeIns(Variant seqChange, Transcript transcript, VariantEffects c
 		returnNow = true; // An insertion can only affect one exon
 	}
 
+	@Override
+	public void codonChange() {
+		// Special case: Is the insertion at the edge of the CDS?
+		int pos = variant.getEnd();
+		if (pos == transcript.getCdsStart() || pos == transcript.getCdsEnd()) {
+			codonChangeCdsEdge();
+			return;
+		}
+
+		// OK, use the 'normal' codon change processing
+		super.codonChange();
+	}
+
 	/**
 	 * Analyze insertions in this transcript.
 	 * Add changeEffect to 'changeEffect'
@@ -74,6 +87,58 @@ protected boolean codonChange(Exon exon) {
 		return true;
 	}
 
+	/**
+	 * Variant (insertion) starting right before CDS' left side
+	 * or right after CDS's right side?
+	 *
+	 * E.g.:
+	 *      [M  P  D  E  E  M  D  D  P  N  P  A ...
+	 *       ^Insertion here just before start codon
+	 *
+	 *       ... P  D  E  E  M  D  D  P  N  P  *]
+	 *                                          ^Insertion here right after stop codon
+	 *
+	 * In either case, the insertion has no coding effect.
+	 */
+	protected void codonChangeCdsEdge() {
+		// Is the insertion at the edge of the CDS?
+		int cdsStart = transcript.getCdsStart();
+		int cdsEnd = transcript.getCdsEnd();
+		int cdsLeft = Math.min(cdsStart, cdsEnd);
+		int cdsRight = Math.max(cdsStart, cdsEnd);
+
+		int pos = variant.getStart(); // Insertions have coordinated 'start = end'
+
+		Exon ex = transcript.findExon(pos);
+		if (pos == cdsLeft) {
+			// Insertion on CDS' left side
+			if (transcript.isStrandPlus()) {
+				// Left side = CDS start
+				codonStartNum = 0;
+				effect(ex, EffectType.FRAME_SHIFT_BEFORE_CDS_START, false);
+			} else {
+				// Left side = CDS end
+				codonStartNum = transcript.protein().length();
+				effect(ex, EffectType.FRAME_SHIFT_AFTER_CDS_END, false);
+			}
+			return;
+		} else if (pos == cdsRight) {
+			// Insertion on CDS' right side
+			if (transcript.isStrandPlus()) {
+				// Right side = CDS end
+				codonStartNum = transcript.protein().length();
+				effect(ex, EffectType.FRAME_SHIFT_AFTER_CDS_END, false);
+			} else {
+				// Right side = CDS start
+				codonStartNum = 0;
+				effect(ex, EffectType.FRAME_SHIFT_BEFORE_CDS_START, false);
+			}
+			return;
+		}
+
+		throw new RuntimeException("This should never happen!");
+	}
+
 	/**
 	 * Get new (modified) codons
 	 */
@@ -94,5 +159,4 @@ protected String codonsAlt() {
 
 		return codonsNew;
 	}
-
 }

src/main/java/org/snpeff/snpEffect/EffectType.java

@@ -76,6 +76,8 @@ public enum EffectType {
 
 	// Modifiers
 	// Order: Highest impact first
+	, FRAME_SHIFT_BEFORE_CDS_START(EffectImpact.MODIFIER) //
+	, FRAME_SHIFT_AFTER_CDS_END(EffectImpact.MODIFIER) //
 	, UTR_5_PRIME(EffectImpact.MODIFIER) //
 	, UTR_3_PRIME(EffectImpact.MODIFIER) //
 	, REGULATION(EffectImpact.MODIFIER) //
@@ -100,14 +102,10 @@ public enum EffectType {
 	, NONE(EffectImpact.MODIFIER) //
 	;
 
-	private final EffectImpact effectImpact;
-
-	private EffectType(EffectImpact effectImpact) {
-		this.effectImpact = effectImpact;
-	}
-
 	static Map<String, EffectType> so2efftype = new HashMap<>();
 
+	private final EffectImpact effectImpact;
+
 	/**
 	 * Parse a string to an EffectType
 	 */
@@ -150,6 +148,7 @@ static void so2efftype(EffFormatVersion formatVersion, Variant variant) {
 		oldSo2efftype.put("inframe_insertion", EffectType.CODON_INSERTION);
 		oldSo2efftype.put("inframe_deletion", EffectType.CODON_DELETION);
 		oldSo2efftype.put("transcript", EffectType.TRANSCRIPT);
+		oldSo2efftype.put("non_canonical_start_codon", EffectType.SYNONYMOUS_START);
 
 		// Add terms if not already in the map
 		for (String so : oldSo2efftype.keySet()) {
@@ -158,6 +157,10 @@ static void so2efftype(EffFormatVersion formatVersion, Variant variant) {
 
 	}
 
+	private EffectType(EffectImpact effectImpact) {
+		this.effectImpact = effectImpact;
+	}
+
 	/**
 	 * Return effect impact
 	 */
@@ -234,6 +237,8 @@ public EffectType getGeneRegion() {
 		case SYNONYMOUS_CODING:
 		case SYNONYMOUS_START:
 		case FRAME_SHIFT:
+		case FRAME_SHIFT_AFTER_CDS_END:
+		case FRAME_SHIFT_BEFORE_CDS_START:
 		case CODON_CHANGE:
 		case CODON_INSERTION:
 		case CODON_CHANGE_PLUS_CODON_INSERTION:
@@ -353,6 +358,12 @@ public String toSequenceOntology(EffFormatVersion formatVersion, Variant variant
 		case FRAME_SHIFT:
 			return "frameshift_variant";
 
+		case FRAME_SHIFT_BEFORE_CDS_START:
+			return "start_retained_variant";
+
+		case FRAME_SHIFT_AFTER_CDS_END:
+			return "stop_retained_variant";
+
 		case GENE:
 			return "gene_variant";
 
@@ -459,7 +470,7 @@ public String toSequenceOntology(EffFormatVersion formatVersion, Variant variant
 			return "stop_retained_variant";
 
 		case SYNONYMOUS_START:
-			return "initiator_codon_variant" + formatVersion.separator() + "non_canonical_start_codon";
+			return "start_retained_variant";
 
 		case TRANSCRIPT:
 			return "non_coding_transcript_variant";

src/main/java/org/snpeff/snpEffect/HgvsProtein.java

@@ -1,5 +1,6 @@
 package org.snpeff.snpEffect;
 
+import org.snpeff.codons.CodonTable;
 import org.snpeff.interval.Transcript;
 import org.snpeff.interval.VariantBnd;
 import org.snpeff.util.Gpr;
@@ -216,7 +217,10 @@ protected boolean isDuplication() {
 	 * Is this a frameShift variant?
 	 */
 	boolean isFs() {
-		return variantEffect.hasEffectType(EffectType.FRAME_SHIFT);
+		return variantEffect.hasEffectType(EffectType.FRAME_SHIFT) //
+				|| variantEffect.hasEffectType(EffectType.FRAME_SHIFT_BEFORE_CDS_START) //
+				|| variantEffect.hasEffectType(EffectType.FRAME_SHIFT_AFTER_CDS_END) //
+		;
 	}
 
 	/**
@@ -337,7 +341,8 @@ protected String posFs() {
 		// Sanity check: Longer than protein?
 		Transcript tr = variantEffect.getTranscript();
 		String protSeq = tr.protein();
-		if (codonNum >= protSeq.length()) return null;
+		if (codonNum > protSeq.length()) return null;
+		if (codonNum == protSeq.length()) return aaCode(CodonTable.TERMINATION_CODON_1) + codonNum;
 
 		// NOTE: the changes observed should be described on protein level and not try to
 		//       incorporate any knowledge regarding the change at DNA-level.

src/main/java/org/snpeff/snpEffect/VariantEffect.java

@@ -514,7 +514,7 @@ public String getEffectTypeString(boolean useSeqOntology, boolean useFirstEffect
 		Collections.sort(effectTypes);
 
 		// More than one effect? Check for repeats
-		Set<String> added = ((effectTypes.size() > 1) && (!useFirstEffect) ? new HashSet<String>() : null);
+		Set<String> added = ((effectTypes.size() > 1) && (!useFirstEffect) ? new HashSet<>() : null);
 
 		// Create string
 		for (EffectType et : effectTypes) {

src/main/java/org/snpeff/snpEffect/testCases/integration/TestCasesIntegrationHgvsFrameShift.java

@@ -57,4 +57,13 @@ public void test_05_hgvs_frameshift() {
 		checkHgvs("testHg19Chr19", path("hgvs_frameshifts_syn_chr19.vcf"), 2);
 	}
 
+	/**
+	 * Test HGVS frameshift at CDS end
+	 */
+	@Test
+	public void test_06_hgvs_frameshift() {
+		Gpr.debug("Test");
+		checkHgvs("testHg3775Chr2", path("hgvs_frameshifts_cds_end.vcf"), 6);
+	}
+
 }

src/main/java/org/snpeff/snpEffect/testCases/integration/TestCasesIntegrationZzz.java

@@ -0,0 +1,12 @@
+package org.snpeff.snpEffect.testCases.integration;
+
+/**
+ * Test case
+ */
+public class TestCasesIntegrationZzz extends TestCasesIntegrationBase {
+
+	public TestCasesIntegrationZzz() {
+		super();
+	}
+
+}

test_04_TestCasesEff.csv

@@ -3,8 +3,8 @@
 
 Name , Value
 Genome , testHg3770Chr22 
-Date , 2018-02-08 11:01
-SnpEff_version , SnpEff 4.3u (build 2018-02-08 10:57), by Pablo Cingolani
+Date , 2018-03-26 14:18
+SnpEff_version , SnpEff 4.3u (build 2018-03-26 14:13), by Pablo Cingolani
 Command_line_arguments , SnpEff  -csvStats test_04_TestCasesEff.csv testHg3770Chr22 tests/integration/eff/eff_sort.vcf 
 Warnings , 1185 
 Number_of_lines_in_input_file, 1205

tests/integration/hgvsFrameShift/hgvs_frameshifts_cds_end.vcf

@@ -0,0 +1,7 @@
+#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO
+2	37480319	.	A	AT	.	.	TR=ENST00000234179;HGVSP=p.Ter891fs;HGVSC=c.2673dupA
+2	37480320	.	T	TA	.	.	TR=ENST00000234179;HGVSP=p.Ter891fs;HGVSC=c.2672_2673insT
+2	37480321	.	T	TA	.	.	TR=ENST00000234179;HGVSP=p.Ter891fs;HGVSC=c.2671dupT
+2	37480322	.	A	AT	.	.	TR=ENST00000234179;HGVSP=p.Ter891fs;HGVSC=c.2670_2671insA
+2	37480323	.	A	AT	.	.	TR=ENST00000234179;HGVSP=p.Ter891fs;HGVSC=c.2669_2670insA
+2	37480324	.	G	GA	.	.	TR=ENST00000234179;HGVSP=p.Pro890fs;HGVSC=c.2668_2669insT

tests/integration/zzz/hgvs_frameshifts_cds_end.vcf

@@ -0,0 +1,7 @@
+#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO
+2	37480319	.	A	AT	.	.	TR=ENST00000234179;HGVSP=p.Ter891fs;HGVSC=c.2673dupA
+2	37480320	.	T	TA	.	.	TR=ENST00000234179;HGVSP=p.Ter891fs;HGVSC=c.2672_2673insT
+2	37480321	.	T	TA	.	.	TR=ENST00000234179;HGVSP=p.Ter891fs;HGVSC=c.2671dupT
+2	37480322	.	A	AT	.	.	TR=ENST00000234179;HGVSP=p.Ter891fs;HGVSC=c.2670_2671insA
+2	37480323	.	A	AT	.	.	TR=ENST00000234179;HGVSP=p.Ter891fs;HGVSC=c.2669_2670insA
+2	37480324	.	G	GA	.	.	TR=ENST00000234179;HGVSP=p.Pro890fs;HGVSC=c.2668_2669insT