From 74ae10fa596b295fdd0904cf0568a3b3d7c5e0a9 Mon Sep 17 00:00:00 2001
From: Tom Clegg <tom@curii.com>
Date: Tue, 30 Nov 2021 14:59:45 -0500
Subject: [PATCH] Call SNPs separately when called within 1bp of start/end of
 indels.

fixes #18496

Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
---
 hgvs/diff.go      | 44 ++++++++++++++++++++++++++++++++++++++++++++
 hgvs/diff_test.go | 15 +++++++++++++++
 2 files changed, 59 insertions(+)

diff --git a/hgvs/diff.go b/hgvs/diff.go
index 835fcc2b97..57025e3588 100644
--- a/hgvs/diff.go
+++ b/hgvs/diff.go
@@ -143,6 +143,50 @@ func cleanup(in []diffmatchpatch.Diff) (out []diffmatchpatch.Diff) {
 			in[i+1].Text+in[i+2].Text == in[i+2].Text+in[i+1].Text {
 			in[i+2], in[i+1] = in[i+1], in[i+2]
 		}
+		// when diffmatchpatch says [delAAA, insXAY] and
+		// len(X)==1, we prefer to treat the A>X as a snp.
+		if i < len(in)-1 &&
+			d.Type == diffmatchpatch.DiffDelete &&
+			in[i+1].Type == diffmatchpatch.DiffInsert &&
+			len(d.Text) > 2 &&
+			len(in[i+1].Text) > 2 &&
+			d.Text[1] == in[i+1].Text[1] {
+			eqend := 2
+			for ; eqend < len(d.Text) && eqend < len(in[i+1].Text) && d.Text[eqend] == in[i+1].Text[eqend]; eqend++ {
+			}
+			out = append(out,
+				diffmatchpatch.Diff{diffmatchpatch.DiffDelete, d.Text[:1]},
+				diffmatchpatch.Diff{diffmatchpatch.DiffInsert, in[i+1].Text[:1]},
+				diffmatchpatch.Diff{diffmatchpatch.DiffEqual, d.Text[1:eqend]})
+			in[i].Text, in[i+1].Text = in[i].Text[eqend:], in[i+1].Text[eqend:]
+			i--
+			continue
+		}
+		// when diffmatchpatch says [delAAA, insXaY] and
+		// len(Y)==1, we prefer to treat the A>Y as a snp.
+		if i < len(in)-1 &&
+			d.Type == diffmatchpatch.DiffDelete &&
+			in[i+1].Type == diffmatchpatch.DiffInsert &&
+			len(d.Text) > 2 &&
+			len(in[i+1].Text) > 2 &&
+			d.Text[len(d.Text)-2] == in[i+1].Text[len(in[i+1].Text)-2] {
+			// eqstart will be the number of equal chars
+			// before the terminal snp, plus 1 for the snp
+			// itself. Example, for [delAAAA, insTTAAG],
+			// eqstart will be 3.
+			eqstart := 2
+			for ; eqstart < len(d.Text) && eqstart < len(in[i+1].Text) && d.Text[len(d.Text)-eqstart] == in[i+1].Text[len(in[i+1].Text)-eqstart]; eqstart++ {
+			}
+			eqstart--
+			out = append(out,
+				diffmatchpatch.Diff{diffmatchpatch.DiffDelete, d.Text[:len(d.Text)-eqstart]},
+				diffmatchpatch.Diff{diffmatchpatch.DiffInsert, in[i+1].Text[:len(in[i+1].Text)-eqstart]},
+				diffmatchpatch.Diff{diffmatchpatch.DiffEqual, d.Text[len(d.Text)-eqstart : len(d.Text)-1]},
+				diffmatchpatch.Diff{diffmatchpatch.DiffDelete, d.Text[len(d.Text)-1:]},
+				diffmatchpatch.Diff{diffmatchpatch.DiffInsert, in[i+1].Text[len(in[i+1].Text)-1:]})
+			i++
+			continue
+		}
 		out = append(out, d)
 	}
 	return
diff --git a/hgvs/diff_test.go b/hgvs/diff_test.go
index cc211b5f2d..1032f4e3d2 100644
--- a/hgvs/diff_test.go
+++ b/hgvs/diff_test.go
@@ -113,6 +113,21 @@ func (s *diffSuite) TestDiff(c *check.C) {
 			b:      "tcaAaagac",
 			expect: []string{"4G>A"},
 		},
+		{
+			a:      "tcagatggac",
+			b:      "tcaAaCggac",
+			expect: []string{"4G>A", "6T>C"},
+		},
+		{
+			a:      "tcagatggac",
+			b:      "tcaAaCggTc",
+			expect: []string{"4G>A", "6T>C", "9A>T"},
+		},
+		{
+			a:      "tcagatggac",
+			b:      "tcaAaCCggTc",
+			expect: []string{"4G>A", "6delinsCC", "9A>T"},
+		},
 	} {
 		c.Log(trial)
 		var vars []string
-- 
2.30.2