diff --git a/regression/run_nuclnucl_result2msa.sh b/regression/run_nuclnucl_result2msa.sh
new file mode 100755
index 0000000..37393c2
--- /dev/null
+++ b/regression/run_nuclnucl_result2msa.sh
@@ -0,0 +1,94 @@
+#!/bin/sh -e
+# Regression test: nucleotide search → result2msa WITHOUT backtrace (-a).
+#
+# Tests both plain and padded (makepaddedseqdb + createindex) target databases.
+#
+# This exercises the fix for result2msa where:
+#   1. NucleotideMatrix must be used (not SubstitutionMatrix) for NT databases
+#   2. Backtraces must be recomputed from alignment records that lack them
+#      (e.g. 14-column records from offsetalignment in the blastn workflow)
+#
+# Without the fix, this produces "DUMMY" all-gap MSA rows or segfaults.
+
+# -- clean previous run --
+rm -rf "${RESULTS:?}"/*
+
+# -- create inline nucleotide FASTA files --
+cat > "${RESULTS}/target.fasta" <<'EOF'
+>seq1 identical_to_query
+CTGCAGCTTGCCCTCAGAGACCGATCTCTCAGAGAGGTACATGGAATCGTGTTCCATCCCTGGATAACGGAACTCTCAGTCCTGCAG
+>seq2 three_snps
+CTGCAGCTTGCCCTCATAGACCGATCTCTCAGAGAGGTACATCGAATCGTGTTCCATCCATGGATAACGGAACTCTCAGTCCTGCAG
+>seq3 short_deletion
+CTGCAGCTTGCCCTCAGAGACCGATCTCTCAGAGAGCGTGTTCCATCCCTGGATAACGGAACTCTCAGTCCTGCAG
+EOF
+
+cat > "${RESULTS}/query.fasta" <<'EOF'
+>query test_sequence
+CTGCAGCTTGCCCTCAGAGACCGATCTCTCAGAGAGGTACATGGAATCGTGTTCCATCCCTGGATAACGGAACTCTCAGTCCTGCAG
+EOF
+
+# -- build databases --
+"${MMSEQS}" createdb "${RESULTS}/target.fasta" "${RESULTS}/targetdb" --dbtype 2
+"${MMSEQS}" createdb "${RESULTS}/query.fasta"  "${RESULTS}/querydb"  --dbtype 2
+
+# padded + indexed variant
+"${MMSEQS}" makepaddedseqdb "${RESULTS}/targetdb" "${RESULTS}/targetdb_padded"
+"${MMSEQS}" createindex "${RESULTS}/targetdb_padded" "${RESULTS}/tmp_idx" \
+    --remove-tmp-files 1 --split 1 --index-subset 0 --search-type 3
+
+ERR=0
+
+validate_a3m() {
+    local label=$1 a3m=$2
+
+    if [ ! -s "$a3m" ]; then
+        echo "FAIL [$label]: result.a3m is empty or missing"
+        ERR=$((ERR + 1))
+    fi
+
+    if grep -q "DUMMY" "$a3m"; then
+        echo "FAIL [$label]: result.a3m contains DUMMY placeholder"
+        ERR=$((ERR + 1))
+    fi
+
+    SEQ_COUNT=$(grep -c "^>" "$a3m" || true)
+    if [ "$SEQ_COUNT" -lt 2 ]; then
+        echo "FAIL [$label]: expected >=2 sequences, got ${SEQ_COUNT}"
+        ERR=$((ERR + 1))
+    fi
+
+    if grep -qE "^-+$" "$a3m"; then
+        echo "FAIL [$label]: result.a3m contains all-gap rows"
+        ERR=$((ERR + 1))
+    fi
+}
+
+# -- Test 1: plain (non-padded) target DB --
+"${MMSEQS}" search "${RESULTS}/querydb" "${RESULTS}/targetdb" \
+    "${RESULTS}/result_plain" "${RESULTS}/tmp" \
+    --search-type 3
+
+"${MMSEQS}" result2msa "${RESULTS}/querydb" "${RESULTS}/targetdb" \
+    "${RESULTS}/result_plain" "${RESULTS}/result_plain.a3m" \
+    --msa-format-mode 6
+
+validate_a3m "plain" "${RESULTS}/result_plain.a3m"
+
+# -- Test 2: padded + indexed target DB --
+rm -rf "${RESULTS}/tmp"
+
+"${MMSEQS}" search "${RESULTS}/querydb" "${RESULTS}/targetdb_padded" \
+    "${RESULTS}/result_padded" "${RESULTS}/tmp" \
+    --search-type 3
+
+"${MMSEQS}" result2msa "${RESULTS}/querydb" "${RESULTS}/targetdb_padded" \
+    "${RESULTS}/result_padded" "${RESULTS}/result_padded.a3m" \
+    --msa-format-mode 6
+
+validate_a3m "padded" "${RESULTS}/result_padded.a3m"
+
+# -- report --
+awk -v actual="$ERR" -v target="0" \
+    'BEGIN { print (actual == target) ? "GOOD" : "BAD"; print "Expected: ", target, "errors"; print "Actual: ", actual, "errors"; }' \
+    > "${RESULTS}.report"
diff --git a/run_regression.sh b/run_regression.sh
index 7b67cf0..27ef325 100755
--- a/run_regression.sh
+++ b/run_regression.sh
@@ -72,6 +72,7 @@ run_test DBPROFILE "run_dbprofile.sh"
 run_test EXPAND "run_expand.sh"
 run_test NUCLPROT_SEARCH "run_nuclprot.sh"
 run_test NUCLNUCL_SEARCH "run_nuclnucl.sh"
+run_test NUCLNUCL_RESULT2MSA "run_nuclnucl_result2msa.sh"
 run_test NUCLNUCL_TRANS_SEARCH "run_nuclnucl_translated.sh"
 run_test CLUSTER "run_cluster.sh"
 run_test EASY_CLUSTER "run_easy_cluster.sh"