Stabilize the obipairing output. whant run twice on the same dataset results are identical

pkg/obialign/pairedendalign.go

@@ -322,9 +322,13 @@ func PEAlign(seqA, seqB *obiseq.BioSequence,
 			}
 			extra5 = -startA
 			startB = 0
+
 			rawSeqA = seqA.Sequence()[startA:]
 			qualSeqA = seqA.Qualities()[startA:]
 			partLen = len(rawSeqA)
+			if partLen > seqB.Len() {
+				partLen = seqB.Len()
+			}
 			rawSeqB = seqB.Sequence()[0:partLen]
 			qualSeqB = seqB.Qualities()[0:partLen]
 			extra3 = seqB.Len() - partLen
@@ -343,6 +347,9 @@ func PEAlign(seqA, seqB *obiseq.BioSequence,
 			rawSeqB = seqB.Sequence()[startB:]
 			qualSeqB = seqB.Qualities()[startB:]
 			partLen = len(rawSeqB)
+			if partLen > seqA.Len() {
+				partLen = seqA.Len()
+			}
 			rawSeqA = seqA.Sequence()[:partLen]
 			qualSeqA = seqA.Qualities()[:partLen]
 			extra3 = partLen - seqA.Len()

pkg/obikmer/encodefourmer.go

@@ -122,6 +122,10 @@ func FastShiftFourMer(index [][]int, seq *obiseq.BioSequence, buffer *[]byte) (i
 		if count > maxcount {
 			maxshift = shift
 			maxcount = count
+		} else {
+			if count == maxcount && shift < maxshift {
+				maxshift = shift
+			}
 		}
 	}
 

pkg/obitools/obipairing/pairing.go

@@ -121,13 +121,14 @@ func AssemblePESequences(seqA, seqB *obiseq.BioSequence,
 	lcons := cons.Len()
 	aliLength := lcons - _Abs(left) - _Abs(right)
 	identity := float64(match) / float64(aliLength)
+	if aliLength == 0 {
+		identity = 0
+	}
+	annot := cons.Annotations()
 
 	if aliLength >= minOverlap && identity >= minIdentity {
-		if withStats {
-			annot := cons.Annotations()
 		annot["mode"] = "alignment"
-			annot["score"] = score
-
+		if withStats {
 			if left < 0 {
 				annot["seq_a_single"] = -left
 				annot["ali_dir"] = "left"
@@ -142,29 +143,28 @@ func AssemblePESequences(seqA, seqB *obiseq.BioSequence,
 			} else {
 				annot["seq_b_single"] = right
 			}
-
-			scoreNorm := float64(0)
-			if aliLength > 0 {
-				scoreNorm = math.Round(float64(match)/float64(aliLength)*1000) / 1000
 		}
-
-			annot["ali_length"] = aliLength
-			annot["seq_ab_match"] = match
-			annot["score_norm"] = scoreNorm
-
 		if inplace {
 			seqA.Recycle()
 			seqB.Recycle()
 		}
-		}
 	} else {
 		cons = JoinPairedSequence(seqA, seqB, inplace)
-
-		if withStats {
-			annot := cons.Annotations()
+		annot = cons.Annotations()
 		annot["mode"] = "join"
 	}
 
+	if withStats {
+		annot["score"] = score
+		scoreNorm := float64(0)
+		if aliLength > 0 {
+			scoreNorm = math.Round(float64(match)/float64(aliLength)*1000) / 1000
+		} else {
+			scoreNorm = 0
+		}
+		annot["ali_length"] = aliLength
+		annot["seq_ab_match"] = match
+		annot["score_norm"] = scoreNorm
 	}
 
 	return cons